Technical ReportPDF Available

Effects of Certified Logging on Wildlife in Community and Industrial Forest Concessions of Northern Guatemala

Authors:

Abstract and Figures

Logging in the Maya Biosphere Reserve is conducted at some of the lowest intensities worldwide (0.8-2.4 trees/ha) and with improved management techniques such as directional felling, road planning, liberation of lianas, and use of lightweight machinery. Three years after timber extraction, we found that ecological impacts of such low-impact harvests are minor and relatively harmless. Several physical impacts of logging were found to be significant. In general, logged areas showed greater canopy openness, lower canopy height, a higher density of seedlings, and a higher density of dead fallen trees than unlogged areas. These structural changes probably drive microclimatic changes, causing logged areas to be warmer and drier than their unlogged counterparts. Several significant faunal responses to logging were found. Of the large vertebrates, only the mantled howler monkey (Alouatta pigra) was found at significantly lower rates in logged areas. Interestingly, no difference was found between logged and unlogged plots in terms of human presence. Almost 70% of the logging impact on game species is explained by timber harvest intensity, suggesting that immediate structural changes are more important determinants of logging impacts on game species than increased access. In general, bird, butterfly, and dung beetle similarity correlated well with logging intensity and/or structural and microclimatic changes caused by logging. Butterfly and beetle communities appear to respond strongly to immediately local changes while bird communities may be more heavily influenced by habitat quality at a wider scale. Community dissimilarity appears to be driven mostly by the addition of new species in logged areas, rather than the exclusion of existing species. For birds and butterflies, logged areas tend to host more species than their unlogged counterparts. The difference is especially marked where logging intensity is high, canopy openness high, and canopy height low. The proportion of intact-forest species in logged plots tends to be equal or higher than the proportion of logged plot species in intact-forest plots. This evidence suggests that increased habitat heterogeneity caused by logging roads and gaps may attract new species, thereby increasing species richness. At current intensities, logging appears not to pose a major threat to the ecological integrity of the Maya Biosphere Reserve. On the contrary, logging operations create jobs for community members, thereby decreasing the likelihood of other, less conservation- friendly land-use practices. The protection of timber and non-timber forest resources also provides an incentive for community members to protect their concessions against forest fires, illegal logging, and illegal colonization. However, if the commercialization of other secondary species increases harvest intensities, impacts may be more severe and should be re-evaluated.
Content may be subject to copyright.
Effects of Certified Logging on Wildlife in
Community and Industrial Forest Concessions of
Northern Guatemala
Jeremy Radachowsky
Rony García, Marcial Cordova, Oscar Aguirre, Ciriaco Marroquin, Tomas
Dubón, Francisco Cordova, Sixto Funes, Juventino López, Gumercindo
García, Francisco Oliva, Gustavo Orellana, Henri Tut, Alejandro
Manzaneros, Emilio Cordova, Pablo Hernandez, Roan Balas McNab
September 2004
MOITORIG ECOLOGICAL ITEGRITY OF THE MAYA
BIOSPHERE RESERVE, PETÉ, GUATEMALA
2
Effects of Certified Logging on Wildlife in Community and Industrial
Forest Concessions of orthern Guatemala
Abstract
Logging in the Maya Biosphere Reserve is conducted at some of the lowest intensities
worldwide (0.8-2.4 trees/ha) and with improved management techniques such as
directional felling, road planning, liberation of lianas, and use of lightweight machinery.
Three years after timber extraction, we found that ecological impacts of such low-impact
harvests are minor and relatively harmless.
Several physical impacts of logging were found to be significant. In general, logged areas
showed greater canopy openness, lower canopy height, a higher density of seedlings, and
a higher density of dead fallen trees than unlogged areas. These structural changes
probably drive microclimatic changes, causing logged areas to be warmer and drier than
their unlogged counterparts.
Several significant faunal responses to logging were found. Of the large vertebrates, only
the mantled howler monkey (Alouatta pigra) was found at significantly lower rates in
logged areas. Interestingly, no difference was found between logged and unlogged plots
in terms of human presence. Almost 70% of the logging impact on game species is
explained by timber harvest intensity, suggesting that immediate structural changes are
more important determinants of logging impacts on game species than increased access.
In general, bird, butterfly, and dung beetle similarity correlated well with logging
intensity and/or structural and microclimatic changes caused by logging. Butterfly and
beetle communities appear to respond strongly to immediately local changes while bird
communities may be more heavily influenced by habitat quality at a wider scale.
Community dissimilarity appears to be driven mostly by the addition of new species in
logged areas, rather than the exclusion of existing species. For birds and butterflies,
logged areas tend to host more species than their unlogged counterparts. The difference is
especially marked where logging intensity is high, canopy openness high, and canopy
height low. The proportion of intact-forest species in logged plots tends to be equal or
higher than the proportion of logged plot species in intact-forest plots. This evidence
suggests that increased habitat heterogeneity caused by logging roads and gaps may
attract new species, thereby increasing species richness.
At current intensities, logging appears not to pose a major threat to the ecological
integrity of the Maya Biosphere Reserve. On the contrary, logging operations create jobs
for community members, thereby decreasing the likelihood of other, less conservation-
friendly land-use practices. The protection of timber and non-timber forest resources also
provides an incentive for community members to protect their concessions against forest
fires, illegal logging, and illegal colonization. However, if the commercialization of other
3
secondary species increases harvest intensities, impacts may be more severe and should
be re-evaluated.
Table of Contents
Introduction ........................................................................................................................ 4
Methods............................................................................................................................... 9
Selection of Treatment Plots......................................................................................... 9
Preparation of Treatment Plots .................................................................................15
Sampling Regimen....................................................................................................... 15
Forest Structure and Composition ............................................................................15
Microclimate ................................................................................................................ 16
Human Presence.......................................................................................................... 16
Large Vertebrates .......................................................................................................16
Birds.............................................................................................................................. 17
Butterflies ..................................................................................................................... 17
Dung Beetles................................................................................................................. 19
Results ............................................................................................................................... 20
Forest Structure........................................................................................................... 20
Microclimate ................................................................................................................ 22
Human Presence.......................................................................................................... 22
Large Vertebrates .......................................................................................................23
Birds.............................................................................................................................. 27
Butterflies ..................................................................................................................... 36
Dung Beetles................................................................................................................. 44
Drivers of Ecological Change in Logged Areas ........................................................ 46
Conclusions ...................................................................................................................... 55
Management Recommendations................................................................................ 56
Recommendations for Research and Monitoring .................................................... 57
Literature Cited................................................................................................................. 58
Appendices ........................................................................................................................ 59
Appendix I. Forest structure summaries by concession ..........................................60
Appendix II. Microclimate summaries by concession ............................................. 64
Appendix III. Bird summaries by concession...........................................................67
4
Appendix IV. Butterfly summaries by concession ................................................... 75
Appendix V. Dung beetle summaries by concession ................................................77
Introduction
Selective logging has been proposed as a potentially sustainable land-use practice in the
Neotropics. Such harvests leave the forest canopy and structure relatively intact, while
providing substantial income to harvesters. Since only a small proportion of Neotropical
forests will ever be maintained in parks and preserves, sustainable forest management is
among the most promising alternatives for protecting large tracts of contiguous forest and
biological diversity. This is especially true in developing countries incapable of financing
and managing parks within a framework of strict protectionism.
In order for sustainable management to be an effective conservation strategy, managed
areas must sustain: 1) viable populations of economically valuable species (timber or
non-timber), and 2) ecological integrity (Whitacre et al. 1992). Managers commonly
monitor the effects of harvests on economically valuable species because their future
income depends upon it. However, ecological side effects of “sustainable” harvests have
been vastly understudied, and no comprehensive investigations had yet been undertaken
in the Northern Neotropics.
Ecological monitoring is a requisite for certification by organizations such as the Forest
Stewardship Council (FSC) and can serve as evidence of responsible forest management,
thereby attracting higher market prices for timber. In some areas, such as High
Conservation Value Forests (HCVF), certification and monitoring may be required by
national legislation (Steve Gretzinger, pers. comm.). Monitoring allows immediate
feedback as to the adequacy of forest management and helps determine the driving
factors behind unacceptable ecological impacts.
This study examines the short-term ecological impacts of certified logging in the Maya
Biosphere Reserve in Northern Guatemala. Because several independent forest managers
with different management practices were logging simultaneously in a relatively
homogeneous area, this study was also able to compare management alternatives and
determine their effects on ecological integrity. Furthermore, this study sets a baseline for
long-term monitoring of logging impacts.
It is important to note that the results herein respond to the impacts of current harvest
intensities. If, as expected, more secondary timber species are harvested and total harvest
intensities increase, the ecological impacts may be more profound. Such potential impacts
should not be extrapolated from the results of this study due to possible threshold or other
unexpected factors, but rather should be investigated anew through a well-designed
monitoring program.
Maya Biosphere Reserve
The Maya Biosphere Reserve (MBR) was established in 1990 in order to protect two
million hectares of subtropical moist forest and savannah in the Petén, Guatemala. The
5
Maya Forest is the largest contiguous tropical forest north of the Amazon and harbors
high levels of biodiversity and endemism. However, this biological reservoir has recently
become imperilled. For over thirty years, the population of the Petén has increased nine
percent per year for a variety of political and socio-economic reasons (Fort and Grandia
1999). Before designation of the reserve, slash-and-burn agriculture and logging
threatened to destroy the entire forest in less than thirty years (Sader 1999). The goal of
the Maya Biosphere Reserve is to prevent such destruction by balancing economic
activity and conservation.
The reserve is divided into three zones (Figure 1). The Core Zone, covering 36% of the
reserve, consists of National Parks and Biotopes. It is reserved for scientific investigation
and low impact tourism. The Multiple Use Zone, covering 40% of the reserve, links the
National Parks and Biotopes. This zone is an 848,440-hectare “extractive reserve” in
which only sustainable, minimally damaging land uses are allowed. The Buffer Zone,
covering 24% of the reserve, forms a band fifteen kilometers wide along the entire
southern border of the reserve.
Figure 1. The Maya Biosphere Reserve, Petén, Guatemala.
In the Maya Biosphere Reserve, the core areas are distributed mainly around the reserve’s
periphery, contrary to the ideal biosphere reserve design. This means that the
Multiple Use Zone must function as the de facto heart of the reserve in terms of
maintaining large-scale ecological processes. The long-term success of the Maya
Biosphere Reserve depends intimately on conservation of the Multiple Use Zone and its
constituent forest concessions.
Forest Concessions in the Maya Biosphere Reserve
As mandated in the "Agreement on Socioeconomic Aspects and the Agrarian Situation"
of Guatemala’s 1996 Peace Accords, extraction rights to timber and non-timber forest
6
products are designated by the Consejo acional de Areas Protegidas (CONAP) through
forest concessions. Currently, there are 14 concessions, ranging from approximately
25,000 hectares to 83,000 hectares, and covering nearly 800,000 hectares in the Multiple
Use Zone of the Maya Biosphere Reserve. All concessions are required by CONAP to
maintain certification, or “green seals”, for timber and non-timber forest product
extraction. Currently all concessions are certified by Smartwood, part of the Forest
Stewardship Council (FSC).
There are several reasons that forest concessions are a potentially viable conservation
strategy in the Maya Biosphere Reserve. First, traditional management is failing. A
variety of factors have coalesced recently in the Maya Forest that place unprecedented
pressure on forest resources. Poverty, ethnic displacement, population increases, special
interests, industrial development, immigration to rural areas, lawlessness, corruption, and
institutional weaknesses have combined to create the highest deforestation rate in Central
America The average rate of deforestation in Southern Mexico and Central America in
the 1980’s was 1.5% per year, while some parts of the Maya Forest were deforested at a
rate of greater than 3% per year (Sader 1999). Laguna del Tigre and Sierra del Lacandón
National Parks are seriously degraded and threatened despite their status.
Furthermore, community forest concessions promise some benefits that open access
harvesting or concessions leased to the highest bidder do not. The first is participation of
local people. Local people often have a unique understanding of the distribution and life-
histories of natural resources, and techniques of sustainable harvesting (Gretzinger 1999).
This is especially true with non-timber forest products. Local people also have a long-
term commitment to the sustainable use of resources within their region. Local people
must ensure that production continues into the future and are more likely to defend their
concessions against exploitation from outsiders. This is true not only because of long-
term commitment, but also because concessionaires are often held responsible for the
quality of their resources (Dugelby 1999). Protected area agencies may fine
concessionaires or rescind their extraction contracts for violations of regulations.
Despite the many benefits of community forest concessions, there are also several
drawbacks. Tamale et al. (1995) listed the following major constraints facing local
participation in forestry: land tenure insecurity, lack of control over forest resources, lack
of reliable markets, lack of appropriate technologies, long rotation periods, competition
with other land uses, and bureaucratic adamancy. Furthermore, internal conflicts within
communities are exacerbated in the Petén as a result of its highly heterogeneous
population, which has been assembled rapidly by recent waves of immigration.
In the Maya Biosphere Reserve, a few of these factors are especially important. Lack of
reliable markets is probably the greatest challenge. Political instability, devaluation of
local currencies, lack of managerial organization and experience, and changing prices on
international markets can cause an economically viable operation to collapse. There is
also a large bias in market demand toward a few valuable species. Furthermore,
communities rarely have the capital to buy equipment such as sawmills and tractors. In
every successfully established community concession in the Maya Biosphere Reserve, at
least one non-governmental organization (NGO) has provided technical and/or financial
7
assistance to the community with international aid. It is still not clear whether the
community concessions will be able to persist without outside support.
Logging Studies in the Maya Biosphere Reserve
Beginning in 1995, the Peregrine Fund started a project entitled “Effects of Logging on
Neotropical Bird and Tree Community Composition” (Schulze and Whitacre 1996).
Using Tikal National Park as an unlogged control and adjacent logged forests north and
south of the park, they compared forest bird community composition. Birds were mist-
netted and surveyed using point counts in each treatment. Of 135 total captured species,
68 were captured in sufficient numbers to allow comparisons of abundances.
Bird community composition was surprisingly similar in logged and unlogged sites. Of
the 135 total species, 97 were found in both forest types. Those that differed were usually
rare species, with less than five individuals caught. Twenty-four species were found only
in the logged forest. All of them have usually been classified as “second growth” species.
Surprisingly, nine “mature forest” species were only found in the logged forest. Three
“second growth” species and ten “mature forest” species were unique to the unlogged
forest.
Comparisons of the 68 common species showed a small, but significant difference. Ten
species showed significant differences between the two treatments. Seven were more
common in the logged forest and three were more abundant in unlogged forest. All of
these species have specific requirements for feeding or nesting, and are probably
responding to structural changes caused by an increased number of gaps in the forest.
In a similar study in the Bethél forest concession, Claudio Méndez compared dung
beetles, butterflies, and small mammals in intact forest, selectively logged forest, and
cattle pastures (1997). He found significant differences between plots in all three taxa.
However, whereas dung beetle diversity and rodent diversity decreased in logged plots,
butterfly diversity increased.
A series of nearly 100 permanent plots have been established in order to measure the
effects of timber harvests on vegetation. The Centro Agronómico Tropical de
Investigación y Enseñanza (CATIE), in conjunction with CONAP, established more than
70 plots (Fundación Naturaleza para la Vida 2000) and Centro Maya established 20 (Sosa
2001). The main focus of this research was to determine the rate of tree regeneration after
timber harvests with different liberation treatments. Clear differences were evident.
Growth rates for all harvestable species were higher after logging (Fundación Naturaleza
para la Vida 2000). This is probably due to increased illumination, which correlated well
with diameter growth rates.
8
Goals of Monitoring the Effects of Timber Extraction
In a participatory threats assessment undertaken in 2001, logging was ranked as the
greatest threat to the ecological integrity of the Maya Biosphere Reserve. However, the
few studies that had examined logging impacts in the reserve were small scale, site-
specific, and gave inconclusive and conflicting results. Furthermore, logging practices
had changed since many of the earlier studies were conducted, and certification had since
become an important factor in forest management. Therefore, an updated and
comprehensive evaluation was necessary.
The study was undertaken in order to achieve the following goals:
Document the direct and indirect impacts of low-intensity timber harvests on
ecological integrity
Determine the effects of different timber management alternatives on ecological
integrity
Develop information to provide forest managers with feedback on current
logging practices
Provide assessments of management practices for certifying organizations
Develop a baseline for long-term monitoring of logging impacts, and for
comparisons with future increases in logging intensity
9
Methods
Selection of Treatment Plots
Although some concessions began selectively logging between 1997 and 1999, 2000 was
the first year in which a sufficient quantity of logged plots were available to allow for
decent replication and across-concession comparisons. Of twelve total concessions that
were logged in 2000, ten were selected for sampling based upon logistic feasibility. The
ten concessions demonstrate a wide variation in management organization and logging
practices (Table 1). Eight concessions are managed by democratic community
organizations and two are managed by private industry.
Table 1. Organization and management of concessions included in study
Concession
Industrial/
community
Organization
name
Community
inside
concession?
Number of
members
Years of
management
before 2000
Arbol verde Community Árbol Verde Outside 344 0
Carmelita Community Carmelita Inside 105 3
Río
Chanchich Community
Impulsores
Suchitecos Outside 27 2
Chosquitán Community
Laborantes del
Bosque Outside 96 0
La Colorada
Community La Colorada Inside 40 0
La Gloria Industrial
Baren
Comercial Outside
N/A
(Industrial) 0
La Pasadita Community APROLAPA Inside 3
Paxbán Industrial GIBOR Outside
N/A
(Industrial) 0
San Andrés Community AFISAP Outside 178 1
Uaxactún Community OMYC Inside 244 0
10
11
Figure 2. Twenty treatment plots used to examine the ecological effects of timber extraction
12
In all of the logged areas of 2000, improved logging techniques such as road planning,
directed felling, and predominant use of lightweight machinery were employed.
However, the methods of road-clearing and post-harvest practices such as road closure
and reforestation (enrichment) were very different (Table 2).
Table 2. Forest management in logged area (POA) of 2000
Concession
Liberation
of lianas?
Planned
roads?
Method of
clearing
roads Felling
Method of
hauling
Road
closed? Reforestation?
Arbol verde
Not
specified Yes
Not
specified Directed Skidder No No
Carmelita
Yes Yes Manually Directed Skidder No
Seed spread
experimentally
Río
Chanchich
Not
specified
Not
specified
Chainsaw,
skidder Directed Skidder
Not
specified Not specified
Chosquitán
Yes Yes Manually Directed Skidder
Branches
placed in
road Not specified
La Colorada
Yes Yes
Not
specified Directed Skidder No Not specified
La Gloria
Yes Yes
Not
specified Directed Skidder
Not
specified
Mahogany
planted
La Pasadita
Not
specified Yes
Chainsaw,
skidder Directed Skidder Closed No
Paxbán
Not
specified Yes
Heavy
machinery Directed Skidder No Not specified
San Andrés
Yes Yes
Not
specified Directed Skidder No No
Uaxactún
Yes Yes Manually Directed Skidder
Branches
placed in
road
Mahogany
planted
13
The size of logged areas in 2000 varied widely between concessions (Table 3). For some
community concessions 2000 was the first year of timber harvests, and therefore they
opted to log small, experimental plots, with as few as 110 hectares. The industrial
concessions logged areas several times as large as those of community concessions. For
example, La Gloria logged 1800 hectares and Paxbán 1450 hectares.
Harvest intensity was greatly limited by the density of marketable species in logged plots
(Table 3). In some areas less than one tree was cut per two hectares, while in others more
than two trees per hectares were extracted. Volumes extracted ranged from less than one
cubic meter per hectare to nearly five cubic meters per hectare, depending upon the
number and the average size of harvested trees.
Table 3. Area and intensity of timber extraction
Concession
Arbol verde
Carmelita
Río Chanchich
Chosquitán
La Colorada
La Gloria
La Pasadita
Paxbán
San Andrés
Uaxactún
mean
Hectares
400
423 390
295 1101800
3381450 800
150 616
# Trees
119
917 883
546 1101139
12117611580
126 730
Volume
257
10341858
1423 2831811
28023511887
336 1152
Trees/ha
0.30
2.17 2.26
1.851.00 0.63
0.36 1.21 1.98
0.84 1.26
Volume/ha
0.64
2.44 4.77
4.822.57 1.01
0.83 1.62 2.36
2.24 2.33
Vol/tree
2.16
1.13 2.10
2.612.57 1.59
2.31 1.34 1.19
2.63 1.96
In total, 25 tree species were harvested in 2000 (Table 4). By far, four commercial
species dominated the harvests: Mahogany (Swietenia macrophylla), Santa María
(Calophyllum brasiliense), Manchiche (Lonchocarpus castilloi), and Spanish Cedar
(Cedrela mexicana). In many cases concessionaires cut fewer trees and species than
those permitted by CONAP due to low market prices at the time of felling.
14
Table 4. Volume of timber extracted per species (m
3
) (includes timber cut and left in patio)
Scientific
Name
Common
Name
Arbol verde
Carmelita
Chanchich
Chosquitán
La Colorada
La Gloria
La Pasadita
Paxbán
San Andrés
Uaxactún
Swietenia
macrophylla
Caoba,
Mahogany 135.0 562.5
569.7
687.1 173.4
1498.6
258.8 1457.7
908.8
118.4
Calophyllum
brasiliense
Santa María,
Marío 103.0
576.0
88.7
33.9
388.2
31.4
34.6
Lonchocarpus
castilloi Manchiche 9.6 324.0
118.1
46.0
44.9
236.5
405.5
36.3
Cedrela
mexicana
Cedro,
Spanish Cedar 7.6
69.8
280.4 42.1
155.2
21.3 121.2
384.8
127.1
Bucida
buceras Pucté
287.0
168.2 34.0
Pseudobombax
elliptica
Amapola
132.5
2.9
79.1
115.3
Vatairea
lundellii Danto
106.8
48.6 67.1
Dendropanax
arboreus Mano de León 6.3
14.4
38.8
16.4
Mastichodendron
foetidesimum Tempisque 1.3
42.4
15.5
Acacia
dolichostachya Jesmó
22.3
16.3
8.5
Aspidosperma
megalocarpon Malerio Blanco
12.3
3.7
6.0
6.9
7.4
Swietenia
panamensis
Chichipate 34.6
Aspidosperma
cruenta Malerio Colorado
16.9
8.7 7.6
Astronium
graveolens Jobillo 12.6
2.9
15.4
Swartzia
lundelli Catalox 1.5
8.5
6.8
5.1
Pouteria
amygdalina Silión
20.7
Terminalia
amazonia Canxán 0.5 1.3
0.4
11.4
Colorín
13.3
Platymiscium
yucataneum Hormigo 7.0
5.3
Rehdera
penninervia
Sacuché
12.1
Pithecellobium
arboreum Cola de Coche
2.2
Metopium
brownei Chechen Negro
0.7
1.2
Pupsikil
1.3
Simira
salvadorensis Saltemuche
1.3
Cordia
sebastena Cericote
0.3
15
Preparation of Treatment Plots
In each logged area, a one square kilometer plot was measured and marked with flagging
tape. Straight-line transects were cut on the entire perimeter and five transects (typically
North-South) were cut at every 200 meters within the square for access to sampling
points. The four corners were marked with large posts, and a section of rebar was driven
flush into the soil so that the plots may be identified precisely with a metal detector even
well after the wooden corner posts have rotted. Schematic maps including the coordinates
of plot corners and the placement of transects are attached as appendices.
During preparation of the logged plot, investigators also characterized slope, soils,
drainage, and vegetation type in order to select an ecologically similar control plot. Using
Landsat images, potential control areas were selected. Priority was given to control areas
close to the harvested area (with at least a 500m buffer), and at similar distances from
communities, water bodies, roads, or other extraneous factors that could affect wildlife
communities. During extensive field visits, habitat types, soils, and slope were verified
for each potential area and the most similar one-kilometer area to the logged plot was
selected. In some cases, topography and vegetation classes forced us to use irregularly
shaped plots in order to maintain comparability with the logged plots. The control plots
were prepared in exactly the same manner as the logged plots.
Sampling Regimen
Investigators worked in two teams of seven people each: two people focused on birds,
forest structure, and microclimate; two people focused on butterflies; two people focused
on dung beetles; and one person served as a cook and guarded the provisional camps and
vehicles during sampling. All investigators working with butterflies and dung beetles also
sampled for large vertebrates in the early morning.
Investigators typically sampled for six to eight days per concession per round, depending
upon logistical conditions and weather. For all taxa, sampling was alternated daily
between the logged and control plots in order to dampen the bias caused by confounding
factors such as weather and temperature. Repeated visits to the same sampling points
allowed calculation of encounter probabilities. Investigators began data collection in
November 2002 and terminated in April 2004, completing four complete rounds of the
ten concessions, and allowing analyses of seasonal and annual trends.
Forest Structure and Composition
In each treatment plot, twenty-five 10 X 10 meter plots were established in order to
measure forest structure data. The plots were positioned in a uniform grid at every 200
meters, always at least 100 meters from treatment plot edges. In each plot, the following
data were recorded: forest type, type of human disturbance, number of standing dead
trees (snags), number of fallen dead trees, average canopy height, maximum canopy
16
height, and presence of fruit on the ground. Leaf litter depth was measured in the four
corners of each plot with a plastic ruler. Percent canopy cover was calculated by taking a
hemispherical digital photo using a Nikon Coolpix 950 camera fitted with a Nikon E-8
fisheye lens, and classifying the images with the program Gap Light Analyzer (GLA),
Version 2.0 (Frazer et al. 1999).
All trees (DBH > 10 cm) within the 10 X 10 meter plot were identified to species.
Diameter at breast height and the presence of fruits and flowers was recorded for each
tree. In a 1 X 10 meter subplot, the number of seedlings and the number of saplings was
also recorded.
A multi-factor MANOVA including treatment type, concession, sampling round, and
sampling team was used to compare continuous quantitative data characterizing habitats
in the two treatments.
Microclimate
At each of the 25 forest structure plots, microclimate data were also recorded. Soil
humidity was measured with a FieldScout TDR 100 soil moisture probe from Spectrum
Technologies with 4.8-inch probes. Soil temperature was measured with a Reotemp brand
soil thermometer with a 4.8-inch probe. Air temperature, relative humidity, and wind
speed were measured with a Kestrel 3000 weather meter.
A multi-factor MANOVA including treatment type, concession, sampling round, and
sampling team was used to compare continuous quantitative data characterizing
microclimate in the two treatments.
Human Presence
All humans and evidence of recent human activity encountered within the 1-square
kilometer study plots was recorded during the entire study period. If possible, the reason
for presence was determined. Dogs were also included as a proxy of human influence. A
paired T-test was used to compare the number of humans in the two treatments.
Large Vertebrates
In each of the ten logged plots and ten control plots, four parallel 1-kilometer straight line
transects were cut at intervals of 200 meters. Transects were cleaned and maintained in
order to allow detection of tracks and other sign, and to avoid making noise while
walking. Flagging tape was placed at every 25 meters along the transects for accurate
distance measurements.
17
Before sampling, all researchers were trained in a participatory two-day workshop in the
Uaxactún forest concession in order to standardize data collection. Investigators practiced
estimating perpendicular distances using flagging tape and studied identifying
characteristics of confusable species.
Observers sampled transects in the early morning between 5:00 and 9:00 AM – the period
in which large vertebrate encounter rates are highest. Observers walked silently at a rate
of 1 km per hour, scanning the forest for all large vertebrates and sign. For each
observation researchers recorded the type of observation (visual, auditory, track, scat,
scratch, fur, feather, or scent), species, number of individuals, position along the transect,
and perpendicular distance from the transect.
To estimate vertebrate densities, conventional distance sampling methods were used.
Visual observations for species with a sufficient number of observations were entered
into the program Distance 4.0 (Thomas et al. 2002). In order to best fit detection
probability curves, data were transformed into intervals and the most parsimonious
detection probability model was selected using Akaike’s Information Criterion (AIC).
Birds
The bird community was sampled using10-minute point counts at the same 25 points at
which forest structure and microclimate data were sampled. In each team, two highly
experienced field assistants sampled 12 or 13 points each daily, recording all visual and
auditory observations in a radius of 100 meters. Hand-held recorders were used to later
verify unknown or confusing vocalizations. Sampling normally began at sunrise, during
peak calling activity, and finished approximately two and a half hours later. All sampling
took place between 5:15 and 10:00 AM.
Community dynamics parameters were estimated through mark-recapture methods using
the program COMDYN (Hines 1999). Species richness was estimated using Burnham
and Overton’s jack-knife estimator (1979). For community similarity comparisons, the
Morisita-Horn index was used.
Butterflies
We employed three methods for butterfly sampling: Van Someron –Rydon baited traps,
visual surveys, and hand netting.
Van Someron –Rydon traps consist of a cylinder of mosquito netting (65 cm high and 25
cm in diameter) supported by two wire loops. The cylinder is completely closed except
for a five centimeter opening at the bottom under which a 40 by 40 cm plywood platform
is suspended. Bait is placed in the center of the platform, attracting butterflies, which
enter through the opening. After feeding, butterflies almost always crawl or fly upward,
and thus become trapped.
18
We placed 25 Van Someron –Rydon traps in each treatment area in a grid at every 200
meters, at a height of 3-5 meters (at the same points used to sample birds, forest structure,
and microclimate). A 200-meter distance, several times the distance used in many other
trapping studies (Austin et al. 1996, Méndez 1997, Sparrow et al. 1994), was used to
assure independence of captures at traps. Through mark-recapture studies with a 100-
meter grid, it was determined that butterflies often traveled from one trap to another in
less time than it took investigators to walk between traps, possibly biasing capture rates
(Radachowsky 2002). Traps were hung from the most suitable tree within 5 meters of the
selected points.
We baited each trap with approximately one cup of mashed rotting bananas mixed with
beer every other day in the morning check after clearing butterflies, alternating sites each
day. Hughes et al. (1998) showed that checking traps after leaving bait out for two days is
more efficient than the typical half-day or one-day periods used in most studies. The bait
was contained in small (ca. 15 cm diameter), open-topped plastic plates in the center of
each trap.
Two people checked one area per day, between 7:00 and 13:00 hours. We removed
specimens one by one from the traps, identifying to species. Individuals not easily
identified in the field were placed in envelopes for later identification. A voucher
specimen of each species and any notable variations within species was collected. For
each specimen, we recorded date, trap location, treatment area, time, and weather
conditions.
During the 5 km walk required to sample traps, butterflies were also identified on the
wing and captured with nets when visual identification was impossible. Investigators
maintained a speed of approximately two kilometers per hour, recording all individuals
observed and caught. For butterflies captured with nets, we kept a voucher specimen of
each species, and recorded date, capture location, treatment area, time, weather
conditions, and whether collected.
Devries’ guides to butterflies of Costa Rica (1997, 1987) and Mariposas Mexicanas (De
La Maza 1986) served as primary identification sources. For its invaluable checklist and
descriptions and photographs of local species, the description of butterflies of the Tikal
vicinity by Austin et al. (1996) was used.
As for birds, community dynamics parameters were estimated through mark-recapture
methods using the program COMDYN (Hines 1999). Species richness was estimated
using Burnham and Overton’s jack-knife estimator (1979). For community similarity
comparisons, the Morisita-Horn index was used.
19
Dung Beetles
To sample dung beetles, we placed 45 baited pitfall traps at intervals of 100 meters along
five parallel transects in each treatment plot. Traps consisted of 10 cm diameter
Tupperware containers buried so that their lips set flush with the soil. Bait was wrapped
in plastic mosquito screening and tied to a stick driven diagonally into the soil so that it
overhung the receptacle. The containers were filled with one inch of soapy water. Traps
were alternately baited with human dung and meat or fish, depending upon availability.
After 24 hours, traps were checked. All beetles captured per trap were placed in a small
bottle with a label describing the date, trap position, treatment type, bait type, and
whether the trap or bait had been disturbed. The same protocol was used for traps with no
beetles.
Specimens were then transported to a central office where they were identified to species
using a stereoscope. A voucher and reference collection was established in the Wildlife
Conservation Society office in Flores, Petén.
The program EstimateS (Colwell 1997) was used to estimate species richness and
similarity indices.
20
Results
In this section, I present results of general interest that describe comparisons between
logged and unlogged treatments and investigate the underlying reasons for ecological
differences. In order to evaluate individual concessions and compare their impacts with
those of other concessions, summaries at the level of individual concessions have been
attached as appendices. Managers and certifying agencies should first read the general
results and conclusions in the main text, and then direct themselves to the corresponding
appendices for individual evaluations.
Forest Structure
Several aspects of forest structure differed significantly between logged and unlogged
treatments (Table 5). In general, logged areas showed significantly greater canopy
openness (p=0.056), a higher density of seedlings, and a higher density of dead fallen
trees than unlogged areas. In unlogged areas, canopy height (both mean and maximum)
was significantly higher. Leaf litter depth, density of dead standing trees (snags), and
density of saplings showed no significant differences at the treatment plot level. Results
of forest structure parameters are summarized by concession in appendix I.
Table 5. Comparison of forest structure parameters in unlogged and logged areas, ordered by the P-value of
a multi-factor manova using concession, round, logging treatment, and team as factors.
Parameter
Control
mean
Logged
mean P(F)
Seedlings (<5cm DBH) 6.6
7.9
0.000
Dead fallen trees 0.8
1.0
0.000
Maximum canopy height (m) 19.2
18.4
0.000
Mean canopy height (m) 13.4
12.9
0.001
Canopy Openness (%) 12.7
13.0
0.056
Leaf litter depth (cm) 2.9
3.0
0.211
Saplings (5-10cm DBH) 0.9
0.9
0.494
Dead standing trees 0.4
0.4
0.729
21
In general, size-class distributions of trees in logged and unlogged treatments are very
similar. However, more trees in the 10 - 20 cm DAP size-class and greater than 100 cm
DAP were encountered in unlogged areas. Mean density is lower in logged areas for the
most heavily harvested size classes (40 cm – 80 cm DAP), although not significantly so.
Table 6. Mean tree density (individuals per hectare) by size-class in unlogged and logged areas.
Differences were tested using a multi-factor manova including concession, round, logging treatment, and
team as factors.
DAP Control
Logged
p(F)
10-20
414.35
376.79 0.000
20-30
138.15
144.29 0.307
30-40
46.24
48.45 0.479
40-50
20.91
17.89 0.125
50-60
9.15
8.32 0.547
60-70
4.24
3.58 0.438
70-80
2.12
1.64 0.427
80-90
1.54
1.55 0.983
90-100
0.19
0.39 0.410
>100
0.67
0.10 0.058
Tree Abundance by Size Class in
Logged Areas and Control
0.01
0.10
1.00
10.00
100.00
1000.00
10-20
20-30
30-40
40-50
50-60
60-70
70-80
80-90
90-100
>100
Individuals per hectare
Control
Logged
Figure 3. Size-class distribution of trees in logged and unlogged areas. Note that densities are presented on
a logarithmic scale.
22
Microclimate
The results of microclimate measurements suggest that logged areas are warmer and drier
than their unlogged counterparts. In logged areas, both air and soil temperature were
significantly higher. Mean air humidity was higher in unlogged plots, though not
significantly. Results of microclimatic parameters are summarized by concession in
appendix II.
Table 7. Comparison of microclimate parameters in unlogged and logged areas, ordered by the P-value of
a multi-factor multi-factor manova using concession, round, logging treatment, and team as factors.
Parameter
Control
mean
Logged
mean P(F)
Air temperature (degrees F) 80.6
81.6
0.003
Soil temperature (degrees F) 74.8
75.1
0.044
Air humidity (%) 78.1
77.0
0.143
Air speed (km/hr) 0.8
0.9
0.200
Soil humidity (%) 44.4
44.4
0.982
Human Presence
In total, 44 humans and 14 dogs were encountered in treatment plots during the study
period. Interestingly, no difference was found between logged and unlogged plots in
terms of human presence (p = 0.87), with 23 people encountered in control plots and 21
people in logged plots.
Table 8. Humans and dogs observed during all sampling sessions
Concession
Humans
Control
Humans
Logged
Total
Humans
Total
Dogs
Distance to
nearest
community
La Colorada
8
210 5
5
Uaxactún 4
711 2
7
La Pasadita 4
610 5
11
Carmelita 0
220
20
San Andrés 2
020
25
La Gloria 5
382
31
Paxbán 0
110
48
Chanchich 0
000
50
Arbol verde 0
000
60
Chosquitán 0 0
0
0 65
Total 23 21
44
14
23
Human presence in both logged and unlogged plots shows similar tendencies with
relation to distance from communities (Figure 4). In plots closer to communities, human
encounter rates were significantly higher than in those far from communities, regardless
of treatment type.
Human Presence in Logged and Unlogged Areas
R
2
= 0.66
R
2
= 0.51
R
2
= 0.7665
0
2
4
6
8
10
12
0 10 20 30 40 50 60 70
Distance to ne arest community (km)
Encounters with humans
Total
Control
Logged
Figure 4. Humans encountered in logged and unlogged areas as a function of distance from communities.
Logarithmic regression analysis shows a similarly strong tendency for higher human encounter rates closer
to human settlements in both logged and unlogged plots.
Large Vertebrates
Encounter Rates and Density
In total, more than 20 species and 5700 observations of large vertebrates were recorded
on the 1087 kilometers of transects sampled during the study period. By far, the most
commonly observed species was the spider monkey (Ateles geoffroyi), followed by
Depp’s squirrel (Sciurus deppei), the crested guan (Penelope purpurascens), chachalaca
(Ortalis vetula), great curassow (Crax rubra), mantled howler monkey (Alouatta pigra),
spotted wood-quail (Odontophorus guttatus), coati (asua narica), great tinamou
(Tinamus major), agouti (Dasyprocta punctata), ocellated turkey (Meleagris ocellata),
brocket deer (Mazama americana), and thicket tinamou (Crypturellus cinnamomeus).
Rank abundance of species encountered is shown in figure 5.
24
Large Vertebrates in Logged and Unlogged Plots
0.00
0.20
0.40
0.60
0.80
1.00
1.20
1.40
Ateles geoffroyi
Sciurus deppei
Penelope purpurascens
Ortalis vetula
Crax rubra
Alouatta pigra
Odontophorus guttatus
Nasua narica
Meleagris ocellata
Dasyprocta punctata
Tinamus major
Crypturellus cinnamomeus
Mazama americana
Tayassu tajacu
Crypturellus soui
Tayassu pecari
Sciurus yucatanensis
Crypturellus boucardi
Odocoileus virginianus
Urocyon cinereoargenteus
Encounter Rate (N/km)
Control
Logged
Figure 5. Rank abundance distribution of large vertebrates sampled with straight line transects
Encounter rates and density point estimates for the 12 most commonly observed species
are given in table 9. Of the more than 20 species of large vertebrates sampled with
transects, only the mantled howler monkey (Alouatta pigra) demonstrated significantly
lower encounter rates in logged areas (p = 0.004). Density estimates for howler monkeys
are 5.67 individuals per km2 in unlogged plots and 3.38 individuals per km2 in logged
areas. Whether density differences are due to the effects of logging or other confounding
factors should be further examined.
25
Table 9. Encounter rates and density estimates for the 12 most commonly observed species sampled with
straight line transects
Species Treatment
Number
of
Obser-
vations
Distance
Sampled
(km)
Encounter
rate (n/km)
D Point
Estimate
(n/km
2
)
Mean
Cluster
Size
Alouatta pigra control
24
528 0.05
5.67 3.47
logged
14
527 0.03
3.38 3.47
Ateles geoffroyi control
213
528 0.40
42.68 3.74
logged
198
527 0.35
37.65 3.74
Crax rubra control
72
528 0.14
4.55 1.45
logged
69
527 0.13
4.13 1.45
Crypturellus cinnamomeus
control
26
528 0.05
2.50 1.02
logged
23
527 0.04
2.13 1.02
Dasyprocta punctata control
26
528 0.05
1.77 1.11
logged
20
527 0.04
1.43 1.11
Mazama americana control
17
528 0.03
0.85 1.05
logged
12
527 0.02
0.57 1.05
Meleagris ocellata control
14
528 0.03
1.65 2.20
logged
12
527 0.02
1.31 2.20
Nasua narica control
14
528 0.03
2.08 2.62
logged
15
527 0.03
2.15 2.62
Ortalis vetula control
64
528 0.12
10.17 2.53
logged
67
527 0.13
11.28 2.53
Penelope purpurascens control
126
528 0.25
12.31 2.18
logged
145
527 0.27
13.71 2.18
Sciurus deppei control
287
528 0.54
31.11 1.14
logged
245
527 0.47
26.54 1.14
Tinamus major control
28
528 0.05
1.82 1.12
logged
25
527 0.05
1.67 1.12
Human Access and Game Species
For the purpose of analyses, the crested guan, great curassow, and brocket deer were
evaluated as indicators for game species. Human access explains nearly 80% of the
variation in game species density in unlogged areas (p = 0.0006) (Figure 6). However,
access shows no significant relationship with game species density in logged areas (p =
0.43) and explains only eight percent of the variance. This suggests that other factors than
access may play a more important role in determining game species densities in logging
treatments.
Figure 7 examines the difference between game species density in paired logged and
unlogged plots as a function of human access. Interestingly, in areas with low human
access, fewer game species were encountered in logged areas than in unlogged areas. In
26
areas with high human access, more game species were found in logged areas than paired
control plots.
Game Species in Logged and Unlogged Areas
R
2
= 0.0807
R
2
= 0.7859
0
5
10
15
20
25
30
35
40
2 4 6 8 10 12
Mean travel time to all communities (hrs)
Game species density (n/km2)
Control
Logged
Figure 6. Game species density in logged and unlogged areas as a function of human access. Note that in
unlogged areas, access explains most of the variation in density while in logged areas other factors are
probably more important.
Game Species in Logged and Unlogged Areas
R
2
= 0.7695
-15
-10
-5
0
5
10
2 4 6 8 10 12
Me an trave l tim e to all comm unities (hrs)
Difference in Game species density
Logged - Unlogged (n/km2)
Figure 7. Difference in game species density between unlogged and logged areas as a function of human
access. Note that near communities differences tend to be positive while far from communities differences
tend to be negative.
27
Birds
Species Richness
In total, 99,713 observations of 224 species were recorded during the study period.
Generally, species richness was greater in logged areas than in unlogged areas (p =
0.059). Only in the concession of La Gloria did species richness in the unlogged plot
surpass that of the logged area. The mean bootstrap richness estimate per concession
across all sampling sessions is shown in figure 8.
Species richness in the logged plots is especially higher than unlogged plots in forest
types that are structurally homogeneous and have few natural disturbances, such as those
sampled in Arbol Verde, Uaxactún, and La Pasadita. Logging may add heterogeneity in
such closed forests, allowing the entry of edge specialists. Species richness summaries by
concession and by sampling round are included in appendix III.
Species Richness in Logged and Unlogged Plots
100
110
120
130
140
150
160
La Colorada
Chosquitán
Paxbán
La Gloria
Carmelita
San Andrés
La Pasadita
Uaxactún
Chanchich
Arbol Verde
Number of species (N)
Control
Logged
Figure 8. Estimated number of bird species in logged and unlogged areas for each concession
28
Similarity of Communities
Phi, the proportion of species in the unlogged plot that were also encountered in the
logged plot, ranged from 0.96 in La Colorada to 0.90 in La Pasadita (Figure 9). Gamma,
the proportion of species in the logged plot that were also encountered in the unlogged
plot, ranged from 0.96 in La Gloria to 0.90 in San Andrés and La Pasadita. That phi and
gamma have similar values suggests no strong directional tendencies in the number of
shared species between treatment plots. If logging were causing the exclusion of species,
one would expect higher gamma values than phi values. Phi and gamma summaries by
concession and by sampling round are included in appendix III.
Bird Species Overlap in Logged and Unlogged Plots
0.88
0.89
0.90
0.91
0.92
0.93
0.94
0.95
0.96
0.97
La Colorada
Uax actún
La Gloria
Chosquitán
Arbol V erde
Carmelita
Chanchich
Paxn
San Andrés
La Pasa dita
Proportion of species (PHI)
Control spp also in
logged a rea
Logged plot spp also in
control
Figure 9. Proportion of species overlap between logged and unlogged plots in ten forest concessions.
The Morisita-Horn similarity index compares both species richness and relative
abundance of species between samples. Encounter rates varied considerably between
field investigators, causing a bias in similarity estimates between teams. Therefore,
similarity indices are calculated separately for both field teams, and ANOVAs also
include sampling team as a factor.
Figure 10 shows community similarity values between logged and unlogged plots for
each team. For team 1, whose encounter rate was significantly higher, similarity indices
ranged from 0.91 to 0.95. For team 2, similarity index values ranged from 0.81 to 0.91.
Interestingly, the rank order of similarity values is very different between teams.
29
Bird Community Similarity in Logged and Unlogged
Plots
0.75
0.8
0.85
0.9
0.95
1
San Andrés
Uax actún
La Pasadita
La Gloria
La Colorada
Carmelita
Arbol Verde
Chanchich
Paxn
Chosquin
Community Similarity (Morisita Horn index)
Team 1
Team 2
Figure 10. Morisita-Horn community similarity between logged and unlogged plots for each sampling
team. Values for teams differ significantly due to a bias in the number of individuals recorded.
Ten bird species were encountered at significantly lower rates in logged areas than in
unlogged areas (Table 10). Of these, nine are resident breeders and one is a winter visitor;
seven are forest generalists, two are edge specialists, and one is a forest interior obligate;
five are omnivores, four are insectivores, and one is a frugivore; five are lower-canopy
foragers, four are aerial salliers, and one is a bark prober; six are understory nesters, two
are cavity nesters, and one is a canopy nester. None of the species are on CITES or IUCN
lists, although Lipaugus unirufus is on the CONAP red list and Lanio aurantius is
endemic to the Maya Forest. It is worth noting that although differences are statistically
significant, declines in encounter rates are mostly less than 25%.
Table 10. Species with significantly lower encounter rates in logged areas according to a multiple-factor
ANOVA with concession, team, round, and treatment as factors.
Scientific Name Common Name Nombre Común
enc rate
control
enc rate
logged p(F)
Columba nigrirostris Short-billed pigeon Paloma piquinegra 0.2 0.17 0.01
Lipaugus unirufus Rufous piha Piha rufa 0.024 0.0157 0.04
Piaya cayana Squirrel cuckoo Cuco ardilla 0.15 0.13 0.03
Mniotilta varia Black-and-white warbler Chipe trepador 0.06 0.04 0.05
Eucometis penicillata Grey-headed tanager Tángara cabecigris 0.18 0.12 0.00
Lanio aurantius Black-throated shrike-tanager Tángara-lanio gorjiinegro 0.25 0.2 0.00
Trogon collaris Collared trogon Trogon collarejo 0.11 0.09 0.01
Trogon violaceus Violaceous trogon Trogon violáceo 0.17 0.15 0.04
Mionectes oleaginus Ochre-bellied flycatcher Mosquero vientre-ocre 0.09 0.06 0.00
Oncostoma cinereigulare Northern bentbill Picocurvo norteño 0.43 0.38 0.00
30
Seven bird species were encountered at significantly higher rates in logged areas than in
unlogged areas (Table 11). Of these, six are resident breeders and one is a winter visitor;
five are forest generalists, and two are edge specialists; three are omnivores, two are
insectivores, one is a frugivore, and one is a seed-eater; three are lower-canopy foragers,
two are upper canopy foragers, one is a ground gleaner, and one is a bark prober; three
are understory nesters, two are cavity nesters, and one is a canopy nester. Aratinga nana
is listed in CITES appendix II and on CONAP’s red list, and Arremonops chloronotus is
endemic to the Maya Forest.
Table 11. Species with significantly higher encounter rates in logged areas according to a multiple-factor
ANOVA with concession, team, round, and treatment as factors.
Scientific Name Common Name Nombre Común
enc rate
control
enc rate
logged p(F)
Aratinga nana astec
Aztec parakeet
Perico pechisucio 0.3 0.46 0.00
Arremonops chloronotus
Green-backed sparrow Gorrión dorsiverde 0.2 0.29 0.00
Cyanocorax yncas Green jay Chara verde 0.01 0.02 0.04
Dumetella carolinensis Grey catbird Pájaro-gato gris 0.25 0.32 0.00
Habia rubica Red-crowned ant-tanager Tángara-hormiguera coronirroja 0.03 0.04 0.01
Melanerpes aurifrons Golden-fronted woodpecker Carpintero frentidorado 0.06 0.08 0.05
Myiopagis viridicata Greenish elaenia Elenia verdosa 0 0.01 0.03
Temporal Variation
More than forty species of migratory birds were sampled in treatment plots. The presence
or absence of these species can temporally affect species richness and community
similarity.
Figure 11 shows altLambda, an estimator for the relative species richness in logged areas
as compared with unlogged areas. Sampling sessions one and four occurred in the winter,
when migratory birds are present in Guatemala, while sessions two and three occurred
during the summer, when there are no migratory birds. Species richness is especially
higher in logged areas during sessions one and four, suggesting a relatively more
important influx of migratory species in logged areas than in unlogged areas. Among
residents, as indicated during sessions two and three, species richness is nearly even
(altLambda close to one) in logged and unlogged plots.
31
Relative Bird Richness in Logged and Unlogged Plots
0.92
0.94
0.96
0.98
1
1.02
1.04
1.06
1.08
1 2 3 4
Sampling Se ssion
Ratio of richness in logged area to
richness in control (altLambda)
Tea m 1
Tea m 2
Total
Figure 11. Relative species richness in logged versus unlogged plots, as estimated by altLambda
When relative abundance is taken into account, as in the Morisita-Horn index,
community similarity appears relatively stable over time. The mean similarity index
value across all ten concessions shows a slight, though insignificant increase over the
study period. This may reflect vegetation regeneration during the one-year study period.
Bird Community Similarity in Logged and Unlogged Plots
0.85
0.86
0.87
0.88
0.89
0.9
0.91
0.92
0.93
0.94
0.95
1 2 3 4
Sampling Session
Team 1
Team 2
Total
Figure 12. Morisita-Horn community similarity index over time
32
Ecological Explanations for Community Differences
Neither the number of individuals nor species of birds showed significant differences
between logged and unlogged areas in terms of primary habitat preference. In logged
areas, two more species of open habitat specialists were encountered, while in unlogged
areas two more generalist species were encountered. The same number of forest-interior
obligate species was encountered in logged and unlogged plots.
Bird Individuals by Primary Habitat Preference
0
5000
10000
15000
20000
25000
30000
35000
Forest Generalist Edge/Open Interior Forest
Obligate
Number of Individuals (N)
Control
Logged
Figure 13. Total number of birds encountered in logged and unlogged plots per habitat specialist guild
Bird Species by Primary Habitat Preference
0
20
40
60
80
100
120
140
Forest Genera list Edge/Ope n Inte rior Fore s t Obligate
Number of Species (S)
Control
Logged
Figure 14. Number of bird species encountered in logged and unlogged plots per habitat specialist guild
33
Interestingly, patterns of bird richness and abundance react differently in logged and
unlogged areas with respect to trophic level (Figures 15, 16). In logged areas, more
insectivorous species but fewer insectivorous individuals were encountered; fewer
frugivorous species but more frugivorous individuals were encountered. To some degree,
these differences may reflect the increase in habitat heterogeneity, but varying responses
in food resource availability in logged areas.
Bird Individuals by Trophic Level
0
5000
10000
15000
20000
25000
Insectivore
Omnivore
Frugivore
Carnivore
Seed-eating
Number of Individuals (N)
Control
Logged
Figure 15. Total number of birds encountered in logged and unlogged plots per trophic level guild
Bird Species by Trophic Level
0
10
20
30
40
50
60
70
80
Insectivore
Omnivore
Frugivore
Carnivore
Seed-eating
Number of Species (S)
Control
Logged
Figure 16. Number of bird species encountered in logged and unlogged plots per trophic level guild
34
Very minor differences were observed in terms of bird foraging behavior in logged and
unlogged plots (Figures 17, 18). In logged plots, more species of bark-probers, soaring
hunters, and lower canopy foragers, and fewer species of ground-gleaners and upper
canopy foragers were encountered.
Bird Individuals by Foraging Behavior
0
2000
4000
6000
8000
10000
12000
14000
16000
18000
20000
Lower-
Canopy
Forage r
Upper-
Canopy
Forage r
Ground
Glea ne r
Aerial
Sa llier
Bark
Probe r
Perched
Hunter
Soaring
Hunter
Number of Individuals (N)
Control
Logged
Figure 17. Total number of birds encountered in logged and unlogged plots per foraging behaviour guild
Bird Species by Foraging Behavior
0
10
20
30
40
50
60
70
80
Lower-
Canopy
Forage r
Ground
Glea ne r
Upper-
Canopy
Forage r
Aerial
Sa llier
Perche d
Hunter
Bark
Probe r
Soaring
Hunter
Number of Species (S)
Control
Logged
Figure 18. Number of bird species encountered in logged and unlogged plots per foraging behaviour guild
35
No significant differences were encountered between logged and unlogged areas in terms
of nest placement guilds (Figures 19, 20). In unlogged areas, four more species of under-
and mid-story nesters were encountered than in logged plots.
Bird Individuals by Nest Placement
0
5000
10000
15000
20000
25000
Under/Midstory
Nester
Cavity Nester
Forest-ground
Nester
Canopy Nester
Generalist
Nester
Open-ground
Nester
Number of Individuals (N)
Control
Logged
Figure 19. Total number of birds encountered in logged and unlogged plots per nest placement guild
Bird Species by Nest Placement
0
10
20
30
40
50
60
70
80
90
Under/Midstory
Nester
Cavity Nester
Canopy Nester
Forest-ground
Nester
Generalist
Nester
Open-ground
Nester
Number of Species (S)
Control
Logged
Figure 20. Number of bird species encountered in logged and unlogged plots per nest placement guild
36
Butterflies
In total, 10,144 individuals of 97 species were observed or captured during the study
period. Of these 9550 individuals and 80 species were caught in baited traps, 582
individuals and 63 species were identified visually, and 12 individuals and 10 species
were caught in nets. Species richness summaries by concession and by sampling round
are included in appendix IV.
Species Richness
In most cases, butterfly species abundance and richness was either higher in logged areas
than in unlogged areas, or nearly equal (Figure 21). Only in Carmelita was butterfly
richness significantly lower in the logged plot than in the unlogged plot. This is due to an
exceptionally high number of species encountered in the Carmelita control plot, and not
an exceptionally low number of species in the logged plot.
Butterfly Species Richness in Logged and Unlogged Plots
0
10
20
30
40
50
60
Carmelita
Chosquitán
La Colorada
San Andrés
Paxbán
Chanchich
La Pasadita
Uaxactún
La Gloria
Arbol Verde
Number of species (N)
Control
Logged
Figure 21. Mean butterfly species richness per sampling session in logged and unlogged paired plots
37
The relative species richness in logged versus unlogged plots, as estimated by altLambda,
demonstrates that logged areas host more species (Figure 22, values > 1). In Arbol Verde,
the logged plot hosted nearly 140% the number of species in the paired control plot.
Again, only in Carmelita were less species encountered in the logged area than in the
unlogged area.
Relative Butterfly Richness in Logged and Unlogged
Plots
0.8
0.9
1
1.1
1.2
1.3
1.4
1.5
Arbol Verde
Chanchich
La Gloria
Chosquin
La Colorada
Paxn
San Andrés
La Pasadita
Uax actún
Carmelita
Ratio of Richness in logged area to richness of
control (altLambda)
Figure 22. Relative species richness in logged versus unlogged plots, as estimated by altLambda
Similarity of Communities
Phi, the proportion of species in the unlogged plot that were also encountered in the
logged plot, ranged from 0.90 in Paxban to 0.68 in Uaxactun (Figure 23). Gamma, the
proportion of species in the logged plot that were also encountered in the unlogged plot,
ranged from 0.90 in Chanchich to 0.58 in La Colorada. In seven out of ten concessions,
phi is greater than gamma, suggesting that the more species rich logged areas may be
adding new species to the community more than excluding species found in uncut plots.
Phi and gamma summaries by concession and by sampling round are included in
appendix IV.
38
Butterfly Species Overlap in Logged and Unlogged
Plots
0.55
0.60
0.65
0.70
0.75
0.80
0.85
0.90
0.95
1.00
Paxbán
Chanchich
Chosquin
Carmelita
La Gloria
Arbol V erde
La Pasa dita
San Andrés
La Colorada
Uax actún
Proportion of species (PHI)
Control spp also in logged are a
Logged plot spp also in control
Figure 23. Proportion of species overlap between logged and unlogged plots in ten forest concessions.
Mean values for the Morisita-Horn similarity index ranged from 0.95 in Uaxactún to 0.60
in Chanchich. Due to anomalous weather conditions and subsequent low capture rates,
some concessions experienced extremely low similarity estimates. It is therefore useful to
examine the maximum similarity index of the four sampling rounds, which ranged from
1.0 in Uaxactún, Chosquitan, and Arbol Verde, to 0.78 in Chanchich.
Butterfly Community Similarity in Logged and
Unlogged Plots
0.55
0.60
0.65
0.70
0.75
0.80
0.85
0.90
0.95
1.00
1.05
Uax actún
Chosquitán
Arbol Verde
La Gloria
La Pasadita
Carmelita
Paxn
La Colorada
San Andrés
Chanchich
Commun ity Similarity (Morisita Horn index)
Maximum
Mean
Figure 24. Maximum and mean butterfly similarity across four sampling rounds
39
Temporal Variation
Figure 25 shows altLambda, an estimator for the relative species richness in logged areas
as compared with unlogged areas. Sampling sessions one and four occurred in the winter,
when most butterflies are inactive, while sessions two and three occurred during the
warm, rainy season, when butterfly richness and abundance are at their peaks. Species
richness is especially higher in logged areas during sessions one and four, suggesting
especially greater activity in logged areas than in unlogged areas during the winter.
Relative Butterfly Richness in Logged and Unlogged Plots
0.6
0.8
1
1.2
1.4
1.6
1.8
1 2 3 4
Sampling Session
Figure 25. Relative species richness in logged versus unlogged plots, as estimated by altLambda
40
Ecological Explanations for Community Differences
Significant differences were found in the number of individuals and species per family
between logged and unlogged treatments. In logged areas, more individuals of the
families Nymphalidae, Pieridae, Papilionidae, and Lycaenidae and more species of
Nymphalids and Pierids were encountered. Pierids, Papilionids, and some Nymphalid
species are open area specialists, only found in clearings, roads, and tree fall gaps.
Structural disturbance caused by logging is likely the cause of increases in these families.
Butterfly Individuals by Family
8891
9999
0
10
20
30
40
50
60
70
80
90
100
Nymphalidae
Hesperiidae
Lycae nidae
Pieridae
Papilionidae
Riodinidae
Number of Individuals (N)
Control
Logged
Figure 26. Number of butterflies per family observed in logged and unlogged areas
Butterflies Species by Family
0
10
20
30
40
50
60
70
80
Nymphalidae Papilionidae Pieridae Lycae nidae
Number of Species (S)
Control
Logged
Figure 27. Number of butterfly species per family observed in logged and unlogged areas
41
In logged areas, more medium and small butterflies were encountered than in unlogged
areas. The number of individuals and species of large butterflies remained constant. This
may reflect the fact that most very large butterflies are forest interior fruit-feeders.
Butterfly Individuals by Size
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
Small Medium Large
Number of Individuals (N)
Control
Logged
Figure 28. Number of butterflies per size class observed in logged and unlogged areas
Butterflies Species by Size
0
10
20
30
40
50
60
Small Medium Large
Number of Species (S)
Control
Logged
Figure 29. Number of butterfly species per size class observed in logged and unlogged areas
42
Butterflies showed slightly different responses in logged and unlogged areas depending
on their color pattern (Figures 30, 31). Most notably, more obliquely-banded individuals
and species were encountered in logged areas, largely due to higher encounter rates of
Opsiphanes cassina. More sulphur-white species and fewer sand-patterned species were
encountered in logged areas.
Butterfly Individuals by Pattern
0
1000
2000
3000
4000
5000
6000
Oblique-band
Closed-reflecting
Canopy-orange
Dark-cryptic
Open-contrasting
Bark
Sand
Black-red
Shrub-orange
Sulfur-white
Adelpha
Black-orange
Tiger
Number of Individuals (N)
Control
Logged
Figure 30. Number of butterflies with distinct patterns observed in logged and unlogged areas
Butterflies Species by Pattern
0
2
4
6
8
10
12
14
16
18
Closed-reflecting
Dark-cryptic
Canopy-orange
Oblique-band
Adelpha
Open-contrasting
Sand
Sulfur-white
Bark
Black-orange
Black-red
Shrub-orange
Tiger
Number of Species (S)
Control
Logged
Figure 31. Number of butterfly species with distinct patterns observed in logged and unlogged areas
43
In logged areas, more butterfly individuals were encountered that use palms (Arecaceae)
as host plants. Although the number of individuals encountered was similar, more species
that use the host plant family Poaceae as their main food source were encountered in
logged areas. This may be due to the proliferation of many species of grasses in roads and
tree fall gaps.
Butterfly Individuals by Host Plant Family
(only families with >500 individuals shown)
0
500
1000
1500
2000
2500
3000
3500
4000
4500
5000
Arecacea e
Euphorbiace ae
Poaceae
Lauracea e
Moraceae
Annonaceae
Number of Individuals (N)
Control
Logged
Figure 32. Number of butterflies by host plant family observed in logged and unlogged areas
Butterflies Species by Host Plant Family
(only families with > 4 butterfly spp. shown)
0
2
4
6
8
10
12
14
16
18
20
Euphorbiaceae
Poaceae
Moraceae
Lauraceae
Arecaceae
Flacourtiaceae
Passifloraceae
Piperaceae
Heliconiaceae
Mimosaceae
Sapindaceae
Caesalpinaceae
Number of Species (S)
Control
Logged
Figure 33. Number of butterfly species by host plant family observed in logged and unlogged areas
44
Dung Beetles
In total, 29,381 individuals of 40 species were captured during the study period. In
unlogged areas, 15,504 individuals of 39 species were trapped. In unlogged areas, 13,807
individuals of 38 species were captured.
Species Richness
In six of the concessions, dung beetle species richness was higher in unlogged areas than
in unlogged areas, while in the other four concessions the pattern is reversed (Figure 34).
Interestingly, where species richness was high in control plots, it tended to be lower in
logged plots.
Dung Beetle Species Richness
in Logged and Unlogged Plots
0
5
10
15
20
25
30
35
Uax actún
Chanchich
La Colorada
San Andrés
Paxn
Arbol Verde
La Gloria
La Pasadita
Chosquitán
Carmelita
Number of species (N)
Control
Logged
Figure 34. Observed number of species in logged and unlogged plots. Note that these are observed data,
and not richness estimates as for birds and butterflies above.
45
Similarity of Communities
Mean values for the Morisita-Horn similarity index ranged from 0.93 in Uaxactún to 0.78
in Chanchich (Figure 35). During the dry season (sampling sessions 1 and 4),
communities in logged and unlogged areas were more dissimilar than in the wet season
(sampling sessions 2 and 3). This may be an effect of low sample sizes and low species
richness estimates in the dry season, when many beetle species are inactive.
Dung Beetle Community Similarity
in Logged and Unlogged Plots
0.50
0.55
0.60
0.65
0.70
0.75
0.80
0.85
0.90
0.95
1.00
Uax actún
Carmelita
Arbol Verde
San Andrés
Paxn
Chosquin
La Colorada
La Gloria
La Pasadita
Chanchich
Community Similarity (Morisita Horn index)
Me an
Maxim um
Minimum
Figure 35. Morisita-Horn index for dung beetle community similarity. Only samples with greater than 5
species were included in figure.
Dung Beetle Community Similar ity in
Logged and Unlogged Plots
0.50
0.55
0.60
0.65
0.70
0.75
0.80
0.85
0.90
0.95
1.00
1 2 3 4
Sampling Session
max
mean
min
Figure 36. Morisita-Horn index for dung beetle community similarity over time.
46
Drivers of Ecological Change in Logged Areas
Above, we examined community differences between logged and unlogged areas for
different animal groups. This section examines the attributes of logged areas that
correlate with those community changes. Please note that correlation does not necessarily
imply causality. In some cases, correlation may be due to covariance between variables.
For example, human access may correlate well with the presence of open-area specialist
butterflies. However, it is not direct human influence that affects butterfly communities,
but rather the forest cover change and forest fires that are often found in areas of high
human access. Correlations should be interpreted carefully with all other evidence and
should take into consideration the life histories of different groups.
Logging Intensity and Forest Structure
Logging intensity shows a significant relationship with several forest structure and
microclimate parameters. For example, nearly 40% of the difference in percentage
canopy openness between paired logged and unlogged plots is explained by the volume
of timber extracted per hectare (p = 0.05) (Figure 37).
Extraction Intensity and Canopy Openness
R
2
= 0.3936
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
0123456
Tim be r volum e e xtracted pe r he ctare (m 3)
Delta Percent canopy openness
Figure 37. Difference in percentage canopy openness between paired logged and unlogged plots as a
function of logging intensity.
47
Nearly 60% of the variation in canopy height differences between paired logged and
unlogged plots is explained by timber harvest intensity (p = 0.01) (Figure 38).
Extraction Inte nsity and Me an Canopy Height
R
2
= 0.5725
-3.0
-2.5
-2.0
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
0 1 2 3 4 5 6
Tim be r volum e e xtracted pe r hectare (m 3)
Delta Mean Canopy Height
Figure 38. Difference in mean canopy height between paired logged and unlogged plots as a function of
logging intensity.
Nearly 50% of the variation in soil temperature differences between paired logged and
unlogged plots is explained by timber harvest intensity (p = 0.02) (Figure 39).
Extraction Intensity and Soil Temperature
R
2
= 0.4987
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
0 1 2 3 4 5 6
Tim be r volum e e xtracted pe r hectare (m 3)
Delta Mean Soil Temperature F
Figure 39. Difference in mean soil temperature between paired logged and unlogged plots as a function of
logging intensity.
48
Logging Intensity and Animal Communities
Logging intensity also demonstrates significant relationships with animal community
changes (Table 12). Game species, birds, and butterflies showed significant trends with
respect to the number of trees extracted per hectare.
Table 12. P-values for linear regression models between timber extraction intensity and faunal similarity
measurements between paired logged and unlogged plots.
Trees
extracted
per hectare
Volume (m
3
)
extracted
per hectare
Game species density
(logged-unlogged)
p=0.03
p=0.00
Howler monkey density
(logged-unlogged)
p=0.65
p=0.30
Birds phi
(% unlogged area
spp in logged area)
p=0.45
p=0.81
Bird similarity
(Morisita-Horn)
p=0.10
p=0.37
Butterflies phi
(% unlogged
area spp in logged area)
p=0.24
p=0.31
Butterfly similarity
(Morisita-Horn)
p=0.09
p=0.16
Beetle similarity
(Morisita-Horn)
p=0.82
p=0.35
Almost 70% of the variation in game species differences between paired logged and
unlogged plots is explained by timber harvest intensity (p = 0.001) (Figure 40).
Logging Intensity and Game Spe cies Density
R2 = 0.689
-15
-10
-5
0
5
10
15
0 1 2 3 4 5 6
Tim be r Volum e Extracte d (m 3)/ Hectar e
Difference in Game Species
Density (Logged-Unlogg ed)
Figure 40. Difference in game species density between paired logged and unlogged plots as a function of
logging intensity.
49
Thirty percent of the variation in bird community similarity between paired logged and
unlogged plots is explained by timber harvest intensity (p = 0.10) (Figure 41).
Logging Intensity and Bird Community Change
R
2
= 0.3005
0.90
0.91
0.91
0.92
0.92
0.93
0.93
0.94
0.94
0.95
0.95
0 0.5 1 1.5 2 2.5
Numbe r of Tre e s Extracted / Hectare
Morisita-Horn Similarity
(Logged-Unlogged)
Figure 41 Difference in bird community similarity (Morisita-Horn index) between paired logged and
unlogged plots as a function of logging intensity.
Slightly more than 30% of the variation in butterfly community similarity between paired
logged and unlogged plots is explained by timber harvest intensity (p = 0.09) (Figure 42).
Logging Intensity and Butterfly Community Change
R
2
= 0.3236
0.75
0.8
0.85
0.9
0.95
1
1.05
0 0.5 1 1.5 2 2.5
Num be r of Tr e e s Extracted / Hectare
Morisita-Horn Similarity
(Logged-Unlogged)
Figure 42. Difference in butterfly community similarity (Morisita-Horn index) between paired logged and
unlogged plots as a function of logging intensity.
50
Forest Structure and Animal Communities
Much of the relationship between logging intensity and faunal responses is probably due
to the structural and microclimatic impacts of logging. Logging may also cause threshold
changes that do not correlate with logging intensity, but exist in any logged area
regardless of the intensity of extraction. This section explores the relationships between
structural and microclimatic changes and subsequent faunal responses.
Two forest structure parameters correlate especially well with logging impacts on fauna:
percent canopy openness and canopy height. Significant relationships exist between these
two parameters and game species, butterfly, and dung beetle community changes (Table
13). Only birds showed no trend with respect to these changes. Birds showed a significant
relationship at the conservative 10% level with change in the number of seedlings (p =
0.09). Where logged areas contained more seedlings than unlogged areas, bird
communities were more dissimilar.
Table 13. P-values for linear regression models between forest structure and microclimate parameters and
faunal similarity measurements between paired logged and unlogged plots.
Game spp
density
change
Bird
similarity
Butterfly
similarity
Beetle
similarity
Canopy openness (%) 0.06
0.18 0.04 0.09
Mean canopy height (m) 0.01
0.50 0.03 0.06
Max canopy height (m) 0.02
0.57 0.02 0.06
Max emergent height (m) 0.10
0.98 0.04 0.15
Seedlings (<5cm DBH) 0.14
0.09 0.01 0.73
Saplings (5-10cm DBH) 0.68
0.79 0.05 0.17
Dead standing trees 0.81
0.99 0.68 0.29
Dead fallen trees 0.27
0.14 0.51 0.87
Leaf litter depth (cm) 0.65
0.85 0.63 0.46
Air temperature (degrees F) 0.72
0.58 0.96 0.26
Air humidity (%) 0.46
0.73 0.53 0.34
Air speed (km/hr) 0.29
0.23 0.75 0.90
Soil temperature (degrees F) 0.11
0.89 0.19 0.74
Soil humidity (%) 0.46
0.63 0.87 0.08
51
Context of Logged Area and Animal Communities
In order to determine the composition of an ecological community, it is often important to
look beyond immediate habitat characteristics and consider a greater context. This may
be especially important in the Maya Biosphere Reserve, where five-year logging plans
often slate harvests contiguous to previously logged areas, thereby increasing the
effective harvested areas several fold. Table 14 shows the p-values for linear regression
models between logging context and fauna.
Table 14. P-values for linear regression models between percent logged area surrounding logged study plot
and faunal similarity measurements between paired logged and unlogged plots.
Percent
logged
area in 1
km buffer
Percent
logged
area in 2
km buffer
Percent
logged
area in 5
km buffer
Game species density
(logged-unlogged)
0.56 0.87 0.65
Howler monkey density
(logged-unlogged)
0.66 0.46 0.85
Birds phi
(% unlogged area
spp in logged area)
0.03 0.05 0.11
Bird similarity
(Morisita-Horn)
0.47 0.66 0.68
Butterflies phi
(% unlogged
area spp in logged area)
0.43 0.32 0.11
Butterfly similarity
(Morisita-Horn)
0.76 0.31 0.89
Beetle similarity
(Morisita-Horn)
0.80 0.73 0.80
52
For birds, logging context predicts phi fairly well (R2 = 0.48, p = 0.03) (Figure 43). This
suggests that the context around a logged area may play a more important role in
determining the proportion of species excluded from logged areas than the parameters of
the logged areas themselves. The number of hectares burned around a logged area also
correlates well with bird community similarity (R2 = 0.60, p = 0.01) (Figure 44).
Logging Context and Bird Community Change
R
2
= 0.4805
0.89
0.9
0.91
0.92
0.93
0.94
0.95
0.96
0.97
0.0 20.0 40.0 60.0 80.0 100.0 120.0
Per centage logged area in 1 k m buffe r of s tudy plot
Proportion of Unlogged Area
Spp in Logged Area
Figure 43. The proportion of species in unlogged areas also found in logged areas as a function of the
amount of logged area surrounding the logged plot.
Forest Fires and Bird Community Change
R
2
= 0.5981
0.89
0.90
0.91
0.92
0.93
0.94
0.95
0 50 100 150 200 250 300
Numbe r of hectare s burned in 5 km buffer of logge d plot
Morisita-Horn Similarity
(Logged-Unlogged)
Figure 44. Bird community similarity as a function of area burnt by forest fires surrounding logged areas
53
Correlation of Faunal Responses
It has often been argued that faunal responses to threats vary so differently that indicator
groups may not indicate much about ecological integrity as a whole. In this section,
faunal responses are compared with one another through correlation and graphical
analysis to examine their utility in making general predictions.
Table 15 gives correlation coefficients between all combinations of indicators. In general,
game species density change best correlates with all other similarity indices (Figure 46).
Morisita-Horn similarity indices also tend to correlate well with each other. Only game
species density change correlates well with howler monkey density change (Figure 45).
Table 15. Correlation coefficients for different faunal indicators
Beetle Similarity
Butterfly
Similarity
Butterfly phi
Bird Similarity
Bird phi
Howler Monkey
change
Game spp
change
Beetle Similarity *
Butterfly Similarity
0.51
*
Butterfly phi -
0.36
-0.09 *
Bird Similarity
0.10
0.44 0.06 *
Bird phi
0.22
0.21 -0.37
0.47
*
Howler Monkey change
0.14
0.01 -0.26
0.11
0.31 *
Game spp change
0.47
0.56 -0.36
0.55
0.24 0.65 *
Game Species and Howler Monkey Density Change
R
2
= 0.4266
-6
-4
-2
0
2
4
6
-15 -10 -5 0 5 10 15
Game spe cies densit y change
Howler monkey density
change
Figure 45. Density change of howler monkeys (Alouatta pigra) as a function of game species density
change
54
Similarity of Faunal Responses to Logging
R
2
= 0.3072
R
2
= 0.3121
R
2
= 0.2202
0.75
0.8
0.85
0.9
0.95
1
1.05
-15 -10 -5 0 5 10 15
Change in game species density
Morisita-Horn Similarity
Butterfly
similarity
Bird
similarity
Beetle
similarity
Figure 46. Butterfly, bird, and dung beetle similarity indices as a function of game species density change
between logged and unlogged areas.
55
Conclusions
Logging in the Maya Biosphere Reserve is conducted at some of the lowest intensities
worldwide and with improved management techniques such as directional felling, road
planning, liberation of lianas, and use of lightweight machinery. If logging can ever be
sustainable and ecologically undisruptive, these are the conditions under which it should
be possible. Three years after timber extraction, we found that ecological impacts of such
low-impact harvests are minor and relatively harmless. Furthermore, logging has
provided incentives for responsible community-based management - a conservation
strategy that may be the best option for maintaining wilderness and wildlife in high-
pressure tropical environments.
Several physical impacts of logging were found to be significant. In general, logged areas
showed greater canopy openness, lower canopy height, a higher density of seedlings, and
a higher density of dead fallen trees than unlogged areas. These structural changes
probably drive microclimatic changes, causing logged areas to be warmer and drier than
their unlogged counterparts.
Several significant faunal responses to logging were found. Of the large vertebrates, only
the mantled howler monkey (Alouatta pigra) was found at significantly lower rates in
logged areas. However, density changes did not correlate with any structural or
contextual parameters of logged areas. Whether density differences are due to the effects
of logging or other confounding factors should be further examined, especially in light of
the species’ recent upgrade from the IUCN status “Least Concern” to “Endangered”.
The secondary effects of logging caused by increased access are often cited as a major
threat to wildlife. Interestingly, no difference was found between logged and unlogged
plots in terms of human presence during this study. This may reflect the forest culture of
Petén, where hunters and non-timber forest product harvesters travel throughout the
forest far from roads. Almost 70% of the logging impact on game species is explained by
timber harvest intensity, suggesting that immediate structural changes are more important
determinants of logging impacts on game species than increased access.
In general, bird, butterfly, and dung beetle similarity correlated well with logging
intensity and/or structural and microclimatic changes caused by logging. Butterfly and
beetle communities appear to respond strongly to immediately local changes while bird
communities may be more heavily influenced by habitat quality at a wider scale.
Community dissimilarity appears to be driven mostly by the addition of new species in
logged areas, rather than the exclusion of existing species. For birds and butterflies,
logged areas tend to host more species than their unlogged counterparts. The difference is
especially marked where logging intensity is high, canopy openness high, and canopy
height low. The proportion of intact-forest species in logged plots tends to be equal or
higher than the proportion of logged plot species in intact-forest plots. This evidence
suggests that increased habitat heterogeneity caused by logging roads and gaps may
attract new species, thereby increasing species richness.
56
At current harvest intensities, it is not surprising that immediate logging impacts do not
exclude forest-interior specialists since harvests typically affect less than 10% of logged
areas. The exclusion of species from logged areas tends to correlate better with the
context of the logged area. For example, in logged areas surrounded by other logged
areas or burnt forest, fewer intact-forest bird species were encountered.
At current intensities, logging appears not to pose a major threat to the ecological
integrity of the Maya Biosphere Reserve. On the contrary, logging operations create jobs
for community members, thereby decreasing the likelihood of other, less conservation-
friendly land-use practices. The protection of timber and non-timber forest resources also
provides an incentive for community members to protect their concessions against forest
fires, illegal logging, and illegal colonization.
However, forest managers should always operate under the precautionary principle due to
incomplete knowledge of tropical ecology, uncertainty in ecological processes, and
potential threshold effects. Many questions remain to be answered. For example, what are
the cumulative effects over time of annual timber harvests? What ecological impacts
would arise if logging intensity were to increase? It is dangerous to extrapolate beyond
the scope of existing data. In order for timber extraction to be sustainable, managers and
certifying organizations must make informed and wise decisions that reflect current
scientific knowledge.
It is important to note that the results herein respond to the impacts of current harvest
intensities. If, as expected, more secondary timber species are harvested and total harvest
intensities increase, the ecological impacts may be more profound. Such potential impacts
should not be extrapolated from the results of this study due to possible threshold or other
unexpected factors, but rather should be investigated anew through a well-designed
monitoring program.
Management Recommendations
Given the above results, we suggest the following management recommendations:
Always minimize canopy opening and road building as much as possible
Close roads after harvests
Use extra precaution in areas inhabited by species of special concern
57
Recommendations for Research and Monitoring
In order to make informed management decisions without sacrificing the profitability of
logging, we suggest the following recommendations for research and monitoring:
Further examine the effects of logging on mantled howler monkey populations
If management techniques remain the same, do not spend resources on repeated
annual evaluations. Medium- to long- term monitoring should be implemented to
examine potential cumulative or threshold effects.
Monitor impacts if logging intensity increases significantly (to > 5m3 / ha)
58
Literature Cited
Burnham, K.P. and W. S. Overton. 1979. Robust estimation of population size when capture probabilities
vary among animals. Ecology 60: 927–936.
Colwell, R. K. 1997. EstimateS: Statistical estimation of species richness and shared species from samples.
Version 5. User’s Guide and application published at: http://viceroy.eeb.uconn.edu/estimates.
Dugelby, B.L. (1999) Chicle extraction in the Maya Biosphere Reserve: a model to estimate sustainability.
Thirteen Ways of Looking at a Tropical Forest: Guatemala´s Maya Biosphere Reserve (ed. by J.D.
Nations), pp. 53-60. Conservation International, Washington, D.C.
Fort, M. & Grandia, L. (1999) Population and environment in the Petén, Guatemala. Thirteen Ways of
Looking at a Tropical Forest: Guatemala´s Maya Biosphere Reserve (ed. by J.D. Nations), pp. 85-91.
Conservation International, Washington, D.C.
Frazer, G.W., Canham, C.D., and Lertzman, K.P. 1999. Gap Light Analyzer (GLA), Version 2.0: Imaging
software to extract canopy structure and gap light transmission indices from true-colour fisheye
photographs, users manual and program documentation. Copyright © 1999: Simon Fraser University,
Burnaby, British Columbia, and the Institute of Ecosystem Studies, Millbrook, New York.
Fundación Naturaleza para la Vida & Organización Manejo y Conservación . Plan Operativo Anual 2000
"El Tanque", Unidad de Manejo Uaxactún, Zona de Usos Múltiples, Reserva de la Biosfera Maya. 1999.
Santa Elena, Petén, Fundación Naturaleza para la Vida.
Gretzinger, S.P. (1999) Concesiones forestales comunitarias: una alternativa económica para la Reserva de
la Biosfera Maya en el Petén, Guatemala. La Selva Maya: Conservación y Desarrollo (ed. by R.B.
Primack, D. Bray, H.A. Galletti and I. Ponciano), pp. 151-165.México.
Hines, J.E., T. Boulinier, J.D. Nichols, J.R. Sauer, and K.H. Pollock. 1999. COMDYN: software to study
the dynamics of animal communities using a capture- recapture approach. Bird Study 46(suppl.), S209-217.
Nichols, J.D., T. Boulinier, J.E. Hines, K.H. Pollock, and J.R. Sauer. Inference Procedures for Animal
Community Dynamics: Changes in Species Richness Over Time and Space. In Prep.
Méndez Hernández, C.A. Diseño de un programa de monitoreo biológico a largo plazo mostrado a través
de un estudio de caso: el corte selectivo del bosque en la cooperativa Bethel, La Libertad, Petén. 1997.
Guatemala, Universidad de San Carlos de Guatemala.
Sader, S. (1999) Deforestation trends in Northern Guatemala: a view from space. Thirteen Ways of Looking
at a Tropical Forest: Guatemala´s Maya Biosphere Reserve (ed. by J.D. Nations), pp. 26-34. Conservation
International Foundation, Washington, D.C.
Schulze, M. & Whitacre, D. Effects of logging on neotropical bird and tree community composition.
1996. Boise, Idaho, The Peregrine Fund.
Sosa Gómez, N.O. Establecimiento y monitoreo de las parcelas permanentes de muestreo bajo la asistencia
de la Asociación Centro Maya. 2001. Petén, Guatemala, CENTRO MAYA.
Tamale, E., Jones, N. & Pswarayi-Riddihough, I. Technologies related to participatory forestry in tropical
and subtropical countries. 1995. Washington, D.C., The World Bank.
Whitacre, D.F., Madrid, J., Marroquin, C., Schulze, M.D., Jones, L., Sutter, J. & Baker, A.J. (1992)
Migrant songbirds, habitat change, and conservation prospects in Northern Petén, Guatemala: some Initial
Results. General Technical Report pp. 339-345.