Professional Documents
Culture Documents
FINAL DRAFT
&
June 2008
Gorongosa National Park
1
TABLE OF CONTENTS
SUMMARY................................................................................................................ 2
1. BACKGROUND & OBJECTIVE ............................................................................ 4
1.1. Original project background............................................................................ 4
1.2. What has changed? ....................................................................................... 5
1.3. Revised objectives & deliverables .................................................................. 5
2. STUDY AREA ....................................................................................................... 7
2.1. Area of Interest............................................................................................... 7
2.2. Description ..................................................................................................... 7
2.3. Data issues .................................................................................................... 9
3. METHODOLOGY................................................................................................ 13
3.1. Available map information ............................................................................ 13
3.1.1. CENACARTA - Carta de Uso e Cobertura da Terra .............................. 13
3.1.2. Tinley..................................................................................................... 13
3.1.3. Cunliffe & Lynam ................................................................................... 13
3.2. Approach to the current study....................................................................... 14
3.3. Field sampling .............................................................................................. 15
3.3.1. Ground sample plots.............................................................................. 15
3.3.2. Aerial and ground control points ............................................................ 15
3.4. Available satellite imagery ............................................................................ 17
3.5. Analysis........................................................................................................ 17
3.5.1. Problems in applying the conceptual approach...................................... 17
3.5.2. Alternative approach.............................................................................. 22
3.6. Data format, storage & availability ................................................................ 22
4. DISCUSSION OF LANDSCAPE CONCEPT ....................................................... 23
4.1. Landscapes, scale and ‘content’................................................................... 23
4.2. Conceptual approach for Gorongosa............................................................ 25
5. REGIONS, LANDSCAPES AND PLANT COMMUNITIES OF GORONGOSA..... 27
5.1. Regions........................................................................................................ 27
5.2. Landscapes.................................................................................................. 29
5.3. Transformed areas ....................................................................................... 35
5.4. Description of landscape delineation and landscape content........................ 37
5.5. Plant communities ........................................................................................ 42
5.5.1. General.................................................................................................. 42
5.5.2. Individual plant communities.................................................................. 49
5.6. Alien plants................................................................................................... 74
6. IMPLICATIONS FOR WILDLIFE ......................................................................... 77
6.1. Introduction .................................................................................................. 77
6.2. Habitat suitability .......................................................................................... 77
6.4. Implications .................................................................................................. 86
7. GAPS & SHORTCOMINGS ................................................................................ 88
8. CONCLUSION .................................................................................................... 90
9. ACKNOWLEDGEMENTS ................................................................................... 90
10. REFERENCES.................................................................................................. 91
APPENDIX A: Glossary of some terms commonly used in the text ......................... 95
APPENDIX B: LISS IV satellite coverage. ............................................................... 97
APPENDIX C: Plant species recorded during this study that are not contained in the
Tinley (1977) list.................................................................................................... 101
APPENDIX D: Structural vegetation terminology (following Edwards (1983))........ 102
SUMMARY
The original objective of the study was to produce a broad-scale landscape map of
the Gorongosa National Park. A landscape is defined as ‘an area with a specific
geomorphology, climate, soil vegetation pattern and associated fauna’. The
landscape unit may encompass several individual plant communities. Yet it reflects a
combination of environmental factors that gives rise to a predictable and repeatable
habitat.
The goalposts changed over time, mainly because the Park is no longer in the same
pioneering phase as some of the other National Parks in Moçambique, such as
Limpopo, Banhine and Zinave, for which landscape maps were drawn up earlier.
Improved satellite imagery was also obtained for Gorongosa. The expectations
therefore shifted towards a more detailed map. The geographical area of interest was
also expanded to encompass the Buffer Zone defined in the Long Term Agreement.
Fieldwork was undertaken. A total of 139 plots were sampled on the ground.
Information was recorded for each plot on the underlying environmental parameters,
vegetation structure, dominant woody and grass species, carrying capacity and
suitability for wildlife. These 139 sample plots are mostly concentrated in the south-
western part of the Park along the few roads and tracks that are easily accessible by
vehicle.
A total of 1,570 control points were assessed during overflights with a helicopter and
during traverses along the main roads in the Buffer Zone. Each of these points has a
GPS locality with information on the structure of the vegetation and where possible
some information on the dominant species. These aerial and ground control points
cover a wider area than the Park, but do not extend in the northern sector to the
Nhamapaza River.
It was originally hoped that a full supervised and/or unsupervised classification1 could
be undertaken of the newly acquired high-resolution LISS IV imagery. That was not
possible. The images were acquired over a two-year period. Underlying differences in
vegetation condition and a large extent of fire scars precluded the ‘stitching’ together of
the individual images. Some limited analysis was done of selected individual images.
A hierarchical approach was used to delineate mapping units. The study area was
split into 4 regions, namely Gorongosa Mountain, Midlands, Rift Valley and
Cheringoma Plateau. These regions correspond to the units originally described by
Tinley (1977). A total of 15 landscapes were defined that fit as follows into the
regions:
1
Some of the technical terms are explained in more detail in Appendix A.
x Midlands Region
The occurrence of alien invasive plant species was recorded during the field survey.
Once a more detailed plant community map has been compiled it will be possible to
predict their potential spread. Probably the potentially most dangerous invader is
Mimosa pigra that occurs on the floodplain landscape.
The landscape map was translated in terms of carrying capacity and suitability for
wildlife. The end product still only provides a broad envelope of potential. This is
because of the problems in delineating the individual plant communities.
The habitat of the Park remains extremely suitable for grazers although the collapse
of the grazing succession has led to a reduction in the extent of the Cynodon-
Dactylon short grasslands which represent a key resource to the herbivores. The
carrying capacity of the Park is high and should allow for stocking rates that are
comparable to historic levels.
Finally, recommendations are provided in order to improve the maps and to address
shortcomings identified during this study.
Following discussions between ICS and the Carr Foundation at the end of 2005 it
was agreed that a base map of the Gorongosa National Park (henceforth GNP) was
required to guide research and management. The base map not only had to provide
units that could be used as a basis for more detailed biodiversity inventories, but it
also had to provide an explanation of the causal factors. The latter is important in that
an understanding of the underlying factors allows for steering, manipulation and
experimentation within the framework of adaptive management.
Vegetation is often used as a surrogate or building block for the definition of habitats.
The use of broad habitat units defined by a combination of environmental factors and
vegetation probably represents the most useful input for further biodiversity surveys.
The proposed approach must take cognisance of the large size of the area, its
relative poor accessibility, the need to provide practically useful input for other
surveys and the requirement of cost-effectiveness.
Experience in the southwest United States indicates that cover-type maps over wide
areas (> 100,000 ha) at reasonable scales (1:100,000 or finer) can take anywhere
from 3 to 5 years to complete with a modicum of accuracy. They can quickly become
expensive (US $ 0.40 - $ 2.00 per ha) (Muldavin et al. 2001). A traditional fine-scale
vegetation description and map are therefore not achievable within a short time
frame and without incurring very significant costs.
At the start of the project it was expected that it would be easy to identify the different
units described by Tinley (1977). These units (probably plant communities) could
then be combined in a number of landscapes. It was anticipated that these
landscapes would be fairly evident and easy to delineate from the high-resolution
LISS IV satellite image that was commissioned. The original ICS proposal in this
regard (December 2005) reads as follows:
Reference was made in the original proposal to the landscape maps produced for the
Limpopo, Banhine and Zinave National Parks (Stalmans 2003; Stalmans et al. 2004;
Stalmans & Wishart 2005).
More than two years have gone by since the original proposal. A number of practical
problems and constraints have been identified:
Since the original proposal there have also been changes in the expectations with
regard to the final product:
x As the Park is no longer in its initial pioneering phase (as was the case with
Limpopo, Banhine and Zinave) the bar has been raised with a more
sophisticated and detailed map being expected by all involved;
x There is a great interest in small-size features such as machamba’s, pans
and dambo’s. These are at a spatial scale that is much smaller than the
landscapes. Somehow these two scales need to be reconciled;
Despite some problems with the Tinley units, it is still felt that Tinley’s (1977) work
provides a hugely important baseline on which any further work should preferably be
based.
It should also be noted that during the course of this study a number of other
outcomes have been achieved that are based on the fieldwork (ground plots and
aerial control points) and analysis that was undertaken:
2. STUDY AREA
Although the GNP forms the core of the Area of Interest for this report, the
landscapes are being assessed for a much wider area (Fig. 1). The total area covers
approximately 10,090 km2 (or 1,009,000 ha). This includes both the GNP (covering
3,688 km2 or 368,800 ha) and its Buffer Zone.
This larger area covers the full geographical extent of the Gorongosa Restoration
Project as per the Agreement for the Long Term Administration of the National Park
of Gorongosa between the Ministry of Tourism of the Republic of Mozambique and
the Gregory C. Carr Foundation. Furthermore, the landscapes on the borders and
upstream of the GNP have a direct bearing on the landscapes within its borders. One
cannot properly understand the landscapes of the GNP without a broader contextual
view beyond its borders.
2.2. Description
The physical environment largely determines soil patterns, vegetation composition and
structure and has direct bearing on land use and development potential. The study area
has been described in much detail by Tinley (1977). Only the most salient and general
features are being summarised here:
x Physiography
o Gorongosa occurs at the southern end of the Great Rift Valley system
that extends from Ethiopia in East Africa down into Moçambique.
o The Rift Valley is the salient feature of the area with its 40 km wide
valley floor that is situated at only 15 to 80 meters above sea level;
o The eastern edge of the Rift Valley rises up to 300 m to form the
Cheringoma Plateau;
o The western edge of the Rift Valley is characterised by the deeply
dissected Midlands region that rises to 400 m;
o The Gorongosa Mountain is perched on the Midlands. It is a massif of
20 by 30 km in size and rises up to 1,863 m above sea level.
x Rainfall
o The Cheringoma Plateau experiences a relatively high rainfall (>1,000
mm). The Rift Valley to the west lies in the plateaus rain shadow and
receives only 700 to 900 mm per annum. The mean annual rainfall at
Chitengo is given as 840 mm (Tinley 1977). The rainfall quickly
escalates with increasing elevation towards Gorongosa Mountain in the
west. Very high rainfall values of over 2,000 mm occur on Gorongosa
mountain. The area immediately north of the mountain lies in its
rainshadow. This is not accurately reflected in the available isohyets (fig.
2).
o The Tinley (1977) isohyets are confirmed by the values obtained from
the 0.5 degree longitude/latitude resolution grid of the Leemans &
Cramer (1991) database.
x Hydrology
o The GNP and its Buffer Zone are drained by a multitude of rivers and
streams. Drainage is from Gorongosa Mountain, the Midlands and the
Cheringoma Plateau down into the Rift Valley. Lake Urema is at the
epicentre of the drainage with the overflow draining into the Pungue
River on its way to the ocean. The eastern boundary of the GNP lies
on the watershed on top of the Cheringoma Plateau. The extreme
eastern part of the Buffer Zone is drained eastwards directly towards
the ocean whereas the western slopes of the Cheringoma plateau are
drained towards the Urema and then the Pungue River.
o Large areas of the Rift valley are regularly inundated. Extreme
floodlevels have been experienced in 2008, for the first time since
1997 (Fig. 2).
x Vegetation
o Broad classifications and maps were produced by Wild & Barbosa
(1967) and White (1983). Tinley (1977) provided detailed descriptions
but no map. A description of the nearby but quite different Catapu
logging concession was recently published (Palgrave et al. 2007).
A recurring theme throughout this study has been the difficulty in matching
information from different sources.
One of the major problems with existing information sources is that the descriptions
and maps apply to much larger areas, often at the scale of the country. As a result,
the boundaries between very different units (e.g. with regard to geology and soils)
may have positional errors of hundreds of meters if not kilometres. Whereas this may
not matter too much at the national or provincial scale, it becomes a real problem at
the scale of the GNP.
Furthermore, the spatial extent of a number of studies does not match. A general
soils map is available for the Sofala province. More detailed information is also
available, but covers only portions of the study area. The soils information at those
different scales does not necessarily match (see Fig. 3).
Waterflow
1863m 300m
273m 370m
597m
& Chitengo
Lake Urema
Gorongosa NP
345m Bufferzone
Geology
Alluvium
Basalt
& 34m Limestone &
Colluvium
Dacite
Elluvium
Eluvium
Gabbro
Gneiss
Granite
& Chitengo Grauwacke
Rift Valley Pebbles
Rivers Quartz
Lake Urema Rhyolite
Gorongosa Mountain & inselbergs Waterflow Tachyte
Gorongosa National Park Weathered rock
Bufferzone N N
20 0 20 40 60 Kilometers 20 0 20 40 60 Kilometers
Nhamapaza
Nhandue
Vundudzi
Nhandare
Muanza
Chitengo
&
Gorongosa NP &
Muaredzi
Bufferzone & Urema
Annual rainfall Pungue
701 - 800
801 - 900
901 - 1000
1001 - 1200 Sapasso
1201 - 1400 & Chitengo
1401 - 1600 Lake Urema
1601 - 1800 Floodwater 2008
1801 - 2000 Rivers
> 2000 Gorongosa Mountain
Gorongosa National Park
N Bufferzone N
20 0 20 40 60 Kilometers 20 0 20 40 60 Kilometers
Fig. 2: Environmental characteristics of Gorongosa National Park and its Buffer Zone.
3. METHODOLOGY
There are three major sources of map information available on the vegetation of the
GNP and its Buffer Zone. These are the CENACARTA landcover maps, Tinley
(1977) and Cunliffe & Lynam (2005).
The CENACARTA maps (Carta de Uso e Cobertura da Terra) depict land cover at a
1;250,000 scale for the whole of Moçambique. The individual units have been
derived through visual identification and manual digitising of units from LANDSAT
satellite imagery. There is virtually no vegetation composition information attached to
the units. Most of the information is in terms of the structural make-up of the
vegetation (namely density and height). Many of the units are also ‘implicit’ in terms
of their information content. This means that within the unit, it is known that other
units occur but their exact position is not depicted. The CENACARTA maps were
evaluated in detail for the Limpopo National Park. The conclusion from that study
was that ‘great care should be used in applying the available landcover data to map
landscapes. The units identified in the landcover undoubtedly make sense, but their
interpretation is unreliable, probably because of insufficient groundtruthing being
possible at a local scale for a project undertaken at the national scale.’ (Stalmans &
Carvalho 2002).
3.1.2. Tinley
Tinley’s (1977) work was aimed at ‘relating the salient features of processes and
correlations into a coevolutionary whole, caught at that particular stage in space and
time by the study’. This attempt at understanding the way in which the landscape is
made up and how the different units relate to each other constitutes a thoroughly
modern approach that goes far beyond a merely descriptive study. The Tinley results
should thus still be relevant in this day and age.
A number of maps are contained in Tinley’s work. They are either at a scale in which
only a few major units are recognised (eg Fig. 6.1. with grasslands, savannas and
forests/thickets) or at a very detailed scale where the intricate mosaic of two or more
plant communities is depicted for a small intensive study area (eg Fig. 9.20).
Tinley did not produce a map at the ‘landscape’ or plant community scale that
covered the whole of the GNP. The reasons for this can only be surmised. It is likely
that the lack of suitable tools, namely a Geographic information system (GIS) and
high-resolution satellite images, were major impediments at that stage.
Cunliffe & Lynam (2005) produced a map that covers most of the Buffer Zone (Fig. 4)
using automated classification of LANDSAT 7 imagery. Some of the drawbacks of
this map are as follows:
x The identified units have very little ‘vegetation content’ attached to it. This is
identified as one of the shortcomings by the authors themselves and they
recommend additional field work;
x There is little evidence of a linkage between the identified units and Tinley’s
descriptions;
x There is extremely limited ground truthing (also identified by the authors) with
the ground truthing not being used to add ‘vegetation content’;
x A large polygon (> 13,000 ha) of ‘escarpment woodland’ (class 18) occurs in
the lower Rift Valley that was inundated during 2008. This is obviously not the
correct classification for the vegetation that is present on the valley floor;
x Some bottomland grasslands and montane grasslands have been lumped
into a single unit (p 13/14 of the report) based on them having the same
spectral signature. However, their relevance to wildlife, their ecological
functioning and requirement management (fire regime) are very different;
Accepting that the Tinley units represent the most detailed and accurate description
that is available at present, the following conceptual approach was proposed at first:
A total of 139 localities were subjectively chosen and assessed (Fig. 5). Due to poor
accessibility, the vast majority of the samples were selected along the roads and
tracks that were accessible at the time of the survey.
The following information was recorded at each sample that covered an area of
approximately 30 m x 30 m:
x GPS position;
x Physiographic unit (Midlands, Rift Valley, …);
x Abiotic factors (landscape position, slope, soil texture, etc.);
x Structure of the woody and of the grasslayer respectively (following Edwards
1983);
x Important woody species (at least 3 major ones);
x Important grass species (at least 3 major ones);
x Overall carrying capacity in ha per Large Stock Unit (estimate based on the
knowledge and experience of the author);
x General suitability for grazers (on a scale of 1 (very low) to 5 (very good));
and
x General suitability for browsers (on a scale of 1 to 5).
A total of 1,570 points were assessed from the air and the ground (Fig. 5). The
quality of information recorded at those points differs from point to point as the
observation time was dependent on the helicopter’s speed and altitude. GPS
positions were recorded together with information on the structure and where
possible the dominant woody species occurring at that spot.
%
%
%
%
%
%
% %
%
%
% %
%
%
% % % %
%
% %
%
% % %% %
% %
% %% % % %%
%% % %% %
%
%
%%%%%%%
% % %% %%
% %%
%%%
%
% % %
%%
% % %% % %% % %
%% %
%% & %
% %% %
%% %%%
%
% %
%
%
% %% %
% %% %
%%% % % % %% %
& Chitengo
% Ground plots
Lake Urema
Gorongosa NP
Bufferzone
N
20 0 20 40 60 Kilometers
##
## #
# # # # # # # # ### #
# ###
#
#
#
## # # ##
#
# # #
# #
# ## #
# # # ##
# # #
## # #
#
## #
#
# # # #
#
#
### # #
# #
#
# #
#
#
# # ## #
## ## # #
# # # #
# ### # #
# #
# # ##
## # ##
#
# ##### # # ## # ####
## # #### # #
### # ## ## #### # # # ##
# # # # # # #
#
####
##
# # ##
# # ##
# # # # ## ##
# # # ###
# # # # # ## #
# # # ##
# # # # # ####### # # # # ##
# # ## # # # ## # #######
# # # ## ### ## ## # #
# ### #
# # ## # ## #
## # # # ## ##
# # ## #
# ## # # # #
# ##
# # # # #
# #
# ## # ## # ### #
#
# # # # #
# # # ## ## #
# ## # # # # # #
## # #
####
# # # ## #
##
# #
# # ## #
# # #
##
# # # ## # ## # # #
### #
# # # #
# # ## ##### ### ## ##
# #
#
#
# #
# #
##
## # # # #
# ## #
#
# ## # # # # ## # # # #
# ##
### ## # ## #
# # # # #
# # # ###
# ## # # #
# # # # #
# # # # #
# # #
# ##
# ## # #####
# #
# # ## # ## # # # # # ## # #
# ## #
## # # ##
#
##
##
# # # # # # ## ##
# ## #
#
### #
#
#
# # ##
#
####
#
## # ## # #
#
#
# ##
# ## # ## #
# #
#
# # # ###
##
#
# # #
# #
# # # # # # ## # # #
###
# ## # #
# ## # ## ## # # ##
#
# ## # ## #
# ##
# # # #
# # # #
## # # # # # # # ## # # # #
# ## # # # # ## # # # # ### #
# ## ## # # # # #
# # ## # # # # ## # ##
# ## # # # ## # # ## #
# ## #####
## # # ## #
# # # ## ## #
# ### # ## #
#
# ## # #
# ## # # # # #
# ### ### #
# ##
# # # ##### ###
# #
###
#
### ## #
# ## # #
##
# # # # # # ###### #### # # # ##### #
# # # #
#
# # ##
## # ## # # # #
# # #
# # # # # # # # # # ## # #
# #
#
# ### # # # ## # # # ## #
#
# # ### # # #
# # # # ## ######
#
# ## ## # # ## # # # # # # ## #
# ## ## ## # # ##
## # # # # # ## ## # ## # #
# # # # # # ## # # # ## ####
# ##
#### #
## #
# # ## ## ## # ### ## # # #### # # # # #
##
# # ## #### ## #### # ## ## ## #
# ### ### # ##
## # ## # # #### # ### #### # ## # ###
# # #
# ## # # ## ## # #
### # # ## # # # # # # # # # # ##
# # ## # # # # ######
# ## # # # ## ## ####### # ## ### ## # # ##
# #
## ## ## # ##### ##
# #
# ## # ## # #
#
#
## ##### ##
#### # ## ##
#
###
#
#####
#
##
#
#
#
# ##
# ###
## # ###
###
#
#
####
# ### #
#
#### &
## ### #
# # ##
#
# #
## #
#
# ##
#
# ##
#
# # #
#
# ### # ## #
#
# ###
## #### ## ## # ###
## # ## # # ## ## ##
# # #
## # # ## # #
### ## #
# ## # # ### ## ## ## ## # ## #
##### #
# # # ### # # ##
# ### #
# ##
# # ### ## # #### ###
## #
## ## ##
## # # #
## # # # # # #
# ### ## #
#
#
### ## # ## ##
## # # # # # # #
# ## #### # #
# ## # #
# # #
#
## # ##
## ##
# ##
## #
& Chitengo
#
Control points
Lake Urema
Gorongosa NP
Bufferzone
N
20 0 20 40 60 Kilometers
Fig. 5: Vegetation samples (2006-2008) in the Gorongosa National Park and its
Buffer Zone.
LANDSAT and ASTER satellite images were available at the start of the study. The
LANDSAT ETM+ (Enhanced Thematic Mapper Plus) image was dated December
2000 and has a resolution of 30 m x 30 m except for the extra 15-meter resolution
panchromatic (=black and white) band. The ASTER scenes were also from the year
2000. They have a resolution of 15 m x 15 m (thus four times better than LANDSAT).
Unfortunately les than half of the study area was covered by ASTER.
During the course of the study high-resolution (5.8 m) multispectral data of the Linear
Imaging Self Scanner (LISS IV) of the Indian Remote Sensing Satellite (IRS) became
available. The LISS IV sensor consists of three spectral bands in the green, red and
near infrared regions of the electromagnetic field.
3.5. Analysis
In practice, the stepwise approach detailed in section 3.2. did not work out as well as
planned. The problems are discussed relative to the five steps that were proposed:
x In practice, it is not always evident to which Tinley unit the vegetation should
be allocated. This problem relates to the lack of quantitative descriptions
identified earlier. There is also an issue of scale. Tinley’s units represent an
idealised and summarized version of the situation as observed at that time
whereas the current field observations incorporate more variability. It may
also be that these units have changed over time (in terms of structure and/or
composition);
x Only a very limited part of the area is easily accessible by vehicle for ground
observations resulting in a very uneven coverage of the GNP;
x Additionally, the field data that were collected during this study were
insufficient in quantity and quality to attempt the ‘normal’ approach of
ordination and classification through which individual units are identified.
Step 3: Use aerial coverage to determine the distribution of those units throughout
the study area;
x The speed of the helicopter and the viewing parallax make that there can be a
positional error of a few pixels between the GPS point and the underlying
image;
x The aerial observations focused more on the structure of the vegetation
(density and height) rather than on its composition;
x The aerial observations can thus not always distinguish between different
units. This means that the aerial coverage operates on a larger scale than the
ground scale.
Step 4: Classify the satellite imagery using automated techniques (supervised and
unsupervised classification) to produce homogeneous units;
x As the high-resolution LISS IV images have been acquired over the course of
two years (Fig. 6), they cannot be ‘stitched’ together and analysed as a single
image. The images have to be analysed one by one and thereafter
mosaiced, which has to do with differing atmospheric conditions and angles at
which the image was captured. This is a huge effort, big enough to have
several students working on these images (Franziska Steinbruch, pers.
comm. 2008);
x There are large fire scars that obscure the underlying ‘true’ nature of the
vegetation (Fig. 7);
x A smoke plume of more than 10 km long and up to 3 km wide on the image
taken on 5 September 2005 obscures some 3,000 ha to the west of Lake
Urema (Fig. 7). The habitat that is not visible represents a transition from
flooded grassland to short grassland that is very important for the grazers;
x It is not known how long the effect of fire scars will ‘persist’. A comparison of
an overlap area between two images (September 2005 and November 2006)
suggest that the effect extends to the next season and that some of the areas
outside the fire scars end up being classified quite differently from year to
year (Fig. 8).
Step 5:Integrate the ground and aerial information, together with ancillary information
on the underlying environmental values, to translate the image units into Tinley units.
&
&
& Chitengo
Lake Urema & Chitengo
Gorongosa NP Lake Urema
Bufferzone Gorongosa NP
LISS IV September 2005 Bufferzone
N LISS IV October 2005 N
20 0 20 40 60 Kilometers 20 0 20 40 60 Kilometers
&
&
& Chitengo
Lake Urema & Chitengo
Gorongosa NP Lake Urema
Gorongosa NP
Bufferzone
Bufferzone
LISS IV September 2006
N LISS IV November 2006 N
20 0 20 40 60 Kilometers 20 0 20 40 60 Kilometers
Fig. 6: LISS IV satellite coverage of the GNP and its Buffer Zone.
Fig. 7: Example of fire scars and smoke plume that obscure the underlying
vegetation (west of Lake Urema in the GNP).
Note: extensive fire scars on Sept 2005 image (classified as red pixels).
Possible persistence of fire scar in Nov 2006 image (classified as blue pixels)
with new fire scars (again classified as red pixels).
This approach relies much more on an a priori, subjective, definition of the desired
outcome followed by a hybrid application of techniques to achieve the outcome. This
approach takes a hierarchical view based on the fact that at a larger scale the
patterns are more general and can be inferred from broad underlying environmental
conditions. At a finer scale, the approach becomes much more driven by the actual
content of the satellite imagery in terms of delineating specific units. However,
because of the problems listed earlier with the imagery, the detailed work is limited to
only a few plant communities rather than the full suite of communities.
The ground and aerial sampling data have been consolidated into an ACCESS
database. There are a total of 971 plant species records for the 139 ground plots and
more than 700 species records for the 1,570 aerial control points. A total of 131 of
the ground plots have associated georeferenced photographs of the vegetation.
Given that individual species were identified during this study (albeit that no
exhaustive sampling was undertaken) it was important to put those in context to the
Tinley information. All species were extracted from Tinley (1977) and consolidated in
a spreadsheet (Stalmans 2006). ‘New’ species that were recorded during the study
and that are not in the Tinley list are presented in Appendix C.
The GIS packages ArcView 3.2 (ESRI 1997) and ArcGis 9.1 were mostly used for
visualisation and analysis. Some use was made of the raster-based IDRISI package
(Eastman 1992) for purposes of manipulation of the remote sensing images. All data
are in the UTM 36S WGS84 projection.
Although the concept of a ‘landscape’ has been defined as ‘an area with a specific
geomorphology, climate, soil vegetation pattern and associated fauna’ there is no
scale attached to the concept.
For obvious reasons, including the costs and workload involved, there is no simple
linear relationship between the size of an area and the number of landscapes that
make up such an area. It is fairly easy for small areas to have a large number of
units that outline in detail the landscape (or even plant communities). However, with
increasing size of the area one tends to have the same number or even less units
that provide a coarser image of the area (Table 1).
Within a specific area, one tends to find that the level of detail goes up with time
elapsed since research efforts have started up. As an example, in the Kruger
National Park, the earliest classifications recognized only a few units. In the last few
decades, the maps became more detailed. The number of individual units was also
dependent on whether the mapping was confined to the Kruger or whether it formed
part of a national program (Table 2).
It is not only the number of units that is important, but it is also the size of the
individual map polygons as this tends to indicate the level of generalization versus
individuality. For example, the average size of the map units in national
classifications of South African vegetation has decreased from 2,313 km2 (Acocks
1953), to 345 km2 (Low & Rebelo 1996) and 71 km2 (Mucina & Rutherford 2006).
The ‘content’ and ‘homogeneity’ of each unit is also important. What exactly is
contained in a defined unit? Gertenbach’s landscapes of the Kruger National Park for
example are heterogeneous and may contain plant communities as different as
woodlands and the seasonal pans found within that woodland (the overriding
requirement is that it must be a repeated and predictable assemblage even though
the components may be very different).
Table 1: Relationship between size of the area and the number of landscapes
identified in southern African national parks.
There are also important differences in terms of the ‘underlying’ information that is
incorporated at the same hierarchical level in several of the landscape maps. As an
example, in Beilfuss et al. (2001) most ‘units’ are based on geology, terrain
morphology and vegetation whereas a specific unit such as ‘coconut plantations’ very
much reflects a land use activity (although obviously such an activity will often only
take place on a unit defined by a specific suitable combination of topography, soils
and rainfall). In a first phase, the same approach was applied by Stalmans to
landscape delineation in the GNP by having different levels of detail at the same
hierarchical level. A ‘mopane’ landscape was for example identified. This is a very
distinct unit that is easy to delineate. However it represents a high level of detail and
high level of homogeneity at a low spatial scale compared to surrounding
landscapes. This approach is thus not consistent.
This approach could also overcome one of the problems with the Cunliffe & Lynam
(2005) map for the GNP. In this map, units smaller than 50 ha in size have been
incorporated into larger units in order to reduce ‘clutter’. This makes sense when
displaying the map at a small scale. However, on large-scale maps one would want
this level of detail, especially as some key elements mostly occur at a scale that is
smaller than 50 ha (dambo’s and pans for example).
The end product is thus a series of maps rather than a single map. It also allows the
‘infill’ of individual landscapes as more detailed plant community maps are compiled
during the course of other research projects. As an example the boundary of the
Limestone Gorge Landscape will (hopefully) not change much. Within it though, a
more detailed appraisal will lead to the mapping of the a number of communities
including the semi-arid edge (Androstachys), the limestone lip and cliffs, and the
riverine forest with the latter probably split up in a number of distinct riverine forest
communities depending on the specific gorge. This concept is further illustrated in
Fig. 9.
It must also be appreciated that the high-level maps represent an idealised image of
a landscape. As an example, the Rift Valley Riverine & Floodplain Landscape is
depicted as a series of polygons. However, within these polygons, there are areas
that are not subject to flooding. Outside these polygons, there are areas that are
flooded from time to time, but they are either too small or they have not been
identified within the available time and resources available for this study.
Fig. 9: Example of increasing level of detail and resolution in vegetation maps for
Gorongosa.
5.1. Regions
The GNP and its Buffer Zone consist of 4 regions as described by Tinley (1977)
(Table 3). They are:
x Gorongosa Mountain;
x Midlands;
x Rift Valley; and
x Cheringoma Plateau.
The Rift Valley occupies the central position of the GNP and its Buffer Zone. The
Cheringoma Plateau region is found to its east and the Midlands to its west. The
Gorongosa Mountain region occupies the central part of the Midlands and is defined
by the 600 m elevation contour (Fig. 10).
The alluvial fans (for example for the Muaredzi River) present a particular case. The
substrate may be similar to that found on the Cheringoma Plateau. However, the fact
that this material gets deposited means that the terrain is flattening out and water
velocity is dropping. Therefore, one is out of the Cheringoma Plateau and into the Rift
Valley. Tinley (1977) used the same reasoning in his delineation of the regions. The
line between the colluvium and the grauwackes on the geological map has thus been
used to a large extent in order to delineate the boundary of the Cheringoma Plateau
region (with some manual adjustments to reflect the actual in-field occurrence of
miombo).
The boundary between the Midlands and the Rift Valley is more ambiguous. There is
a difference on whether one uses the rift faulting or the dominant occurrence of
alluvial material. In the northern part of the area, there are extensive alluvial deposits
along the Nhamapaza River. A cut-off was selected that is located somewhat west of
the faulting, but that only captures the major alluvial area and not the upstream
narrow alluvial belt.
Table 3: Size and relative extent of the different regions in Gorongosa National Park.
EN1 to Caia
to
Inhaminga
&
Rivers
Major roads
& Chitengo
Lake Urema
Gorongosa Nat Park
Gorongosa Mountain Region
Midlands Region N
Riftvalley Region to Dondo
W E
Cheringoma Plateau Region
S
20 0 20 40 60 80 100 Kilometers
5.2. Landscapes
The following 15 landscapes have been identified (Fig. 11)(Plates 1 – 4)(Table 4):
x Midlands Region
The above would provide the level of detail that was expected as per the original ICS
proposal. However, as discussed earlier, the expectations and requirements are now
higher and additional detail needs to be provided. This is done at the next
hierarchical level of the ‘plant communities’.
EN1 to Caia
Rivers
Major roads
Gorongosa N.P.
Transformed areas
Gorongosa Lower Montane Grassland & Woodland
Gorongosa Montane Grassland & Shrub-forest
Gorongosa Montane Forest
Midlands Moist Miombo
Midlands Dry Miombo & Mixed Woodland
Midlands Alluvial
Midlands Inselberg
Riftvalley Alluvial Fan
Riftvalley Riverine & Floodplain
Riftvalley Lake Urema to Dondo
Riftvalley Colluvial Fan
Cheringoma Plateau Seaward Slope
Cheringoma Plateau Calcareous Sandstone
Cheringoma Plateau Argillaceous Sandstone N
Cheringoma Plateau Limestone Gorge
20 0 20 40 60 Kilometers
A: Midlands Moist Miombo Landscape – note steep terrain and rocky riverbeds.
B: Midlands Moist Miombo Landscape – largely cleared for the cultivation of sorghum
C: Midlands Dry Miombo & Mixed Woodland Landscape – in the rainshadow northwest of
Gorongosa Mountain.
D: Midlands Dry Miombo & Mixed Woodland Landscape – mixed woodland without miombo
elements.
E: Midlands Inselberg Landscape with miombo.
F: Midlands Alluvial Landscape – Boa Maria ruins (with miombo) and cultivated areas
A: Rift Valley Alluvial Fan Landscape – dense palm and Piliostigma communities
B: Rift Valley Alluvial Fan Landscape – tall mixed woodland
C: Rift Valley Riverine & Floodplain Landscape – flooded grasslands
D: Rift Valley Riverine & Floodplain Landscape – short grasslands with recent burns
E: Rift Valley Lake Urema Landscape – hippo house with Lake Urema in background
F: Rift Valley Lake Urema Landscape – flooded grassland around Lake Urema.
Table 4: Extent and proportion of the Park and of the study area covered by the
individual landscapes.
Transformed areas have been added to the landscape map. The transformed areas
of Fig. 11 have been delineated using a combination of information sources:
x Cunliffe & Lynam (2005) map (but adapted as some areas on the
Cheringoma Plateau were erroneously identified as transformed (on
account of the bare ground in the naturally eroded areas of the
Argillaceous Sandstone Riftward Slope landscape));
x Land cover map (Carta de Uso & Cobertura) – those polygons that are
designated as having more than 50% settlements and/or cultivation);
x Unsupervised classification of the LISS IV images around Gorongosa
Mountain;
x Visual assessment of the LISS IV images;
x Woods Hole information on recent forest loss on Gorongosa Mountain
(Beilfuss et al. 2007);
x GPS localities of settlements and machamba’s observed during the
helicopter flights.
The transformed areas cover almost 72,500 ha that is spread across different
landscapes (Table 5).
The extent of transformation within the Park is limited to some 2,489 ha (or less than
0.7%). Whereas the absolute extent of the transformation is limited, it is the position
of some of the cultivated areas within the Rift Valley Region that has a relatively large
impact. This impact is direct (in terms of immediate disturbance and potential source
of illegal hunting) as well as indirect (in terms of curtailing restoration potential and
tourism development options).
The landscapes have been delineated and can be shortly described as follows (see
also Plates 1 to 4):
5.5.1. General
It is not possible at this stage to accurately map all of the plant communities that
make up the different landscapes. This is because of limitations in the available data
and available time resources. However, a number of the plant communities are key
to either the functioning of the ecosystem or as resources to the wildlife. Others
reflect important impacts on the landscape (eg machamba’s).
In terms of key resources, the dambo’s, the Cynodon-Digitaria short grasslands, and
the saline grasslands and woodlands are vital for the wildlife (see also section 6.).
The pan communities are key as sources of water and of grazing as water levels
recede.
The aerial observations did not always enable the discrimination between different
grassland or different palmveld communities. For this reason, the observations were
grouped in relatively large ‘community groupings’ (Fig. 12). A cross tabulation was
done between these observations and the Cunliffe & Lynam polygons (Table 6). Only
those cross tabulations where more than 50% of the points within a certain polygon
corresponded to a specific ‘community grouping’ were retained and mapped (Fig.
13).
The cross tabulation indicate a very good match of 75% between the observed
machamba’s and class 26 of Cunliffe & Lynam ‘Cultivation’. This matching
percentage increases to 88% if one considers that a number of secondary and
degraded miombo points fall in the Cunliffe & Lynam ‘cultivation ‘polygons. This is a
very good match for a remote sensing outcome. Classification accuracy reported by
other researchers mapping landcover (Brondizio et al. 1996, Fuller et al. 1998,
Hodgson et al. 1988, Lunetta & Balogh 1999), forest cover (Congalton et al. 1993)
and grassland cover (Lauver & Whistler 1993) ranged from 69 to 94%.
However, only a limited number of other good matches were recorded (Table 6). This
was for Class 19 Barue miombo woodland and Class 20 Planalto miombo woodland
in the miombo, Class 4, 5 and 6 for the grasslands and Class 22 and 24 for the
forests. Some units have a high percentage of the control points all belonging to the
same ‘community grouping’ that is however very different from the Cunliffe & Lynam
determination (see Class 23 Mountain grassland versus ‘miombo grouping’ and
Class 17 Dry thicket / forest versus ‘Acacia – Combretum grouping’).
Although there is a good match between the field data and the ‘classes’, there are
significant errors of omission and commission. Errors of omission occur when the
satellite map fails to recognise the actual habitat class of an independent sample
plot. Errors of commission result when the satellite map misclassifies the sampling
plot from the independent data set. A large area of grassland is omitted in the bottom
left quadrant of the study area (compare Fig. 11 and 12) whereas a large polygon of
Class 18 ‘escarpment open woodland’ (=miombo) supposedly occurs along the
Urema River. Its ‘true’ character is grassland (as evident from the control points).
Even in the absence of a detailed community map, information is still required on the
extent of the communities in order to get a better handle on aspects of carrying
capacity and suitability for wildlife. The control points represent the outcome of a
subjective approach that did not necessarily sample the study area uniformly. There
are for example proportionally more control points for the mopane as the boundary of
the two patches was flown and GPS’ed. Therefore, the use of the proportional
representation of the control points by different communities would result in a biased
estimate of their extent. In order to partially counter this bias, the Park was overlain
by a systematic grid of 3 km by 3 km (Fig. 14). The community grouping was
recorded for each of the control points closest to each of the grid intersections. In this
way, a total of only 382 points were selected from the sample of more than 1,200
points.
Are these results realistic? Tinley (1977) (in his Fig. 6.1) states that the grasslands
occupy 20.8% of the Rift Valley. The current study comes up with a figure of 28.7.6%
for the Rift Valley region in the Park. Tinley’s Rift Valley sector includes the northern
extension towards the Nhamapaza River. Although the Tengane and Nhamasingu
floodplain grasslands occur in that sector, it can be expected that the overall
percentage grassland is lower than that occurring downstream.
# # # #% 0#0
#0 #
#
# W#
'
# W
#0
0 'W' ##
#W' %U
%
%'
W '
W 0
U
W
' U'
W %
'
W0
# ### ' #
0
U'
%
#W #
0 0W
W
'
0
W0
'# '
W
#
0
''
#W
W0
'
W
x
x W
'
#
0
x
W
'
x
0#
#
x #
# U
%
%
U U
% U#
% x U
%
#
0 U%
% #
U #% U #
0
U
%
## # W
'
#
0 #%
U
# W
' #
# ###
# #
0 #
U 0
% # ##
#0#
#
# # #0 #0 #xx x #x # 0
#
0 ####
##0 0 #0#0
###'
0 W0 #' W0
#
0 x #
z #
0 x ' Wx #W
U
% ## ##
# # x x
# # ' #
W
'
# #
0 W
'
W
'
# x
x
x
z #
0
#
0 x
#
x
0
0#
#
0 0x x
# # x
# % #
0#'
0
#
0
W' W' W U
% U
%U
%W'
'
U
%W
#
x
x # 0 # 0#
0##
0 #x #
0
#
00#x 0
x0
# x
#x x % U%
% U ' W' W % U % U%
U%%'
U' U'
%%
W
W
#
0
#x xx
#
0 x x
#
0 % S
# % U
W
'W% '
' W W
'
#
0
W %
#
0 # 0 0
#x0 # #
0 % %' U0
% % W%
'W
' W
'U
% #
0
#
0
# # x # #
# x% #x W
' % W # W
' W
'
U
%
#
0 %
x 0 #
0 # x
#0
0 # U%
U ' % W
'
x # x
#
x x % U
% W %W
' #W
' W
'
x
z%%
x 00 W0 ' W %%% x
x
#
0
#
0 #
0
x x x
x W
'#
%% W
' W%
' U'
U% #
x
x#
#
#
0 #
0 xx
# x
# %% U
% %% U' W W
'#W
' W
' % W
' U
% # x
# x
x #
#0 0
x x W
'
% ' W W'
' W z x%% #
0#
0
#
0 x
# x%
x xx U
% W
' U
% W
' W
' W
'W
' #
0 %
U x
# #
# ##
x xxx
x#xxx x U
% W
' x
z U
% W
' U
% x
#
x
# ###
0x
#xx
0 x W
'
x
z W
' W
'
%
W
' WW
' # % xx
#
0 x xx W xz
' W
' xW
' W
' W
' #
0
#0'
# x x
x 0# xxx
#
xx
# #
% W
' x
zW
'z U
%xz
z
x
z
x
z x
x %%
W
' U U %0
% #
0
#
0#
0W
'x x
x
# x #
x0
x#x
xxx U
% x ' %%% zx x
zx
z U
% U
%U
% ## #
# 0
0 #
#0
0 #
x x# xU
% x
z
# x
z
W
'W%z
' W U S # U
% x
zz
x
zx
z
x
x
zzx
z
xx
z
z
x
zx x %
z
x
z UU
% W
' x
x # %
x
# # x x W
' U
% zx
xzxx
z
U 0
% 0
# #x
0
x#x#
x
# x xx
U
% # %% ' W
U% U% US#%% UU U
% z
z x
z
%x %
z U% U ### x%
xx
x
# W
' zx
# % % % S
#S
#U
% U
%U %
% % U
% U
% W'
' W
x
U
%
# x % U #
%
# #
xxxx
#x
# W %
'
U% ' W # %% z
x ' W% ' W'
x
'W S
U %%
%
#
U
%
U
%
U
%
U
%
U
%
#0 x
# x
x
x 0 # x x #' W ' W
' %
% z
U
% WSS
#
U
%
# U
% % U %
U
%%
U%
UU
%
UU
%
S
# U
%
x
z
U
%
U
%x
zxS
z #
x
z S#% ' W0 0
## # x
% x #0
0# x
WW % z x #% % U %
#S
# U% U%
% U%U
U%
% U%
U UU
% %% W%
' x x x0#xx
x x W'
'
W
' W#% ' Uz
%
#
# x
z
z
x
x
x
z
xz
x
x%
z %%%%
% U
U
%
US
U
%
U
%
%
U%
%
U
U % U%
U
%
U%
U%%
U%
UU
% U%
% U U%
U%U%
x
x
#
0
#
0 U
% % W
' % % x %
x
z
#z
x
z
zx
z
xz U
%
U
% U
%
U
% U U
% U
% # % #
0
# x U ## x%
z U%% U%
%U% U%
U%U%% U
% U%
% x x
x # '
#
W
x
W
'
x %% % U%
% U
U%
%
U U% UU%
% U%
%U%
U
%
U U%
% U
% U
% U
U %
%
U
% U
% US
#% % U'W'WWx
'
#
x
#x
#
x
x 0
# x xx
#
U'
% W
' %
W
'
U
% U
%U
% %U
U%
% U
U
%
% U
% U
% W
' x x
x #
x W %'
% x
#
W% % U%
# U%U% %%'
%
U
%
U
%
W
' zx
W
#S#
% U'W S#%% % %
#%' W %
S
# U
%
U%%%
U
%S
#%%
U%
U%%%
U%% U
%%U%
%
UU%
U% UU
%
U
%
U%
%U
#
x x x
x x U
% U
%U
%
UUS
% S# %% % %
U S# U
% x
xx#x %S
#%S%%%%%
# US
%
#
W'
' S
#z%%%%
% %% ' ##SS#S#% 0S
# x
xx xx # x
#x
#W
'x
x#
x
x
x'
W
' x#
W
x
#
#
%WW
'
W
' W
'
W
'
%'W
W'
'W
W
'
%%
x
z
S
U
%#%#S
#'W#'
W W%
'Ux
% %W
%#%S %
S
##%S #S#U
%#%
S
#S
#S
#
US
#S
# U
%
SS
#S
#
# #%
U
S
U
%
U
%
Ux
z #
x
x
x
x
Ux
%
x x x xx x #
x#
#x #x
#
x# W
'W
'
W
'U
%
% W
'
U
% %U
x
%
z
W
'% %%% W
'%W
'W
' %W
' W
' W
'W
' zxS# %
W
' # % W
' S
#S
# S
# S# zx
U
%
W
'
S
#% W
'% x
# x
# x
x
#
#
xx
# x x %
W#'
' W
x x
# x
x
#x
#
xx
#
W
'x %%
W
'
%U
%
U
%U
% U
%
U
%%
%
#
S
#
xW
'
x %
x
S
#x
z U
% x
z
%
#
S%%
%W
' W
' W
' x
z % x
zU
%
S
# x
xxxx x xx x
x W
'
x #x x U
% W
'%%W
'
U
%S
#U
%#z
#%
U
%
S
#Uz
%
W
'S
#Sz
S
S
#
#S#
S
S
##U
%U
%
x
z
xS
#S
# x
zSS
#
x
S
#
z
# SU
%
# x
zW
' W
' S
#% x
zS
#
U
% S
# U
%
U
%
S
#
S#
U
% x x x
x#
x
U
%
x Ux
U%
% #'x#%
W U
%
xx
W%
U U%
%U S
#
z
W
'WS
'
x# #%Uz #'
SWz
x z
Wxz WS
' #
W
' # U'
% WS U %
% U
%
US
# x x
#x
0x#
xx x
x
x W'
' %'
W%'
' WW US
%
U
%
U
%
#
U
%
U
%
U
%
U
%
U
%
S
#S
U
% S
# ' xz
x#S #S
S
##'
#S #W ' W
#
S
#S
W
' #
U
%
S
#
U
%
U%
%U
U
%
S
#U'
% x
x
#
x S
# #x
zSS
# W
' S# W x x
## z xz x%
U% #
U
%U#S
% #
x x x
#
x W%
' U S %%%
U #S# x# x U %x
%
#
U
%##
xx% U S
#%U% U%
WS
' #S
#S#Wz
'
S
#
x#%U%
U x %xx xx
x x
U% S
# x
z x x
UU%
% U z
% W%
'
W
'U% x xx
U% U%U% U% S
#
U%
Ux#x x xx xxxx
U%U x
xxx xx x
Aerial & ground control points U%
%U
U%
% U U
% xxxxx
x
x xx x
W Acacia - Combretum
' xx x
% Acacia communities
zx Broad-leaved communities
U
% Grassland communities
# Forest communities N
#
0 Cultivation / Settlement
# Miombo communities
S
# Palm veld communities
0 10 20 30 Kilometers
Rift Valley Lake Urema
Fig. 12: Plant community ‘groupings’ for the ground and aerial control points in the
Gorongosa National Park.
Table 6: Cross tabulation between Cunliffe & Lynam (2005) classes and ‘community
groupings’ of the aerial and ground control points. The table must be read as follows:
Fig. 13: Cunliffe & Lynam (2005) vegetation ‘classes’ that are ‘translated’ into
‘community groupings’ where the match between aerial / ground control points and
the ‘classes’ exceeds 50%.
# ##
#
#
# # # #
#
# # # ##
# # # #
# # #
#
# ## # # #
##
#
# # #
# # # #
# ## # #
# #
# ## # ## #
# #
# ## # #
#
# # # ## ##
# # ## #
# # ## # #
# # #
# #
# # # # ## #
#
# ## # # # # #
# # #
# ### # #
# # # # # #### ## #
# # # # #
# # # # #
# ## # ## # # ##
# #####
# #
# # # # # #
# # # # # ##
## # # #
# # # #
## #
# # # # #
#
# # # ## ## #
# #
# # ## #
## ##
# # ## ## # #
# # # # #
# ## #
# #
#
# # # # # ## # # # # #
# # # #
###### #
## # # # #
##
# # ### ## # ##
## #
##
# # # #
### # # # ## # #####
# #
#
#
# #
# #
## # #
###
# # ## ##
# #
# # ### ## #
# #
# # # # # # ## # #
# #
# # # ## #
## # # # ## #
# # # # # # #
# ## # #
## # # ## #
# # #
## # # #### ## ## # # # # ###
## # #
# ## # ## ##
##
## ### # # #
# # # ### ## #
# # ## ### ### # # #
## # #
#
# ## # ##
# ### #### ##
##### #
## # #
#
# # # #
## ##
# # #
# # # #
# # # # #
# #
# # # # ##
## ## # ## # ##
# ## # #
## ### ## ## ## # ## # # #
# # # ## # # # # #
# # #
#
# # # # # # ## # # #
# # ## # ## ## # #
# # # ## # # ## ## # ## ###
## ## ## # # ##
## # ##
## # #### ### #### #
# #
#
## # #### # # ###
#
#
# ## # # # # # # # ## ###
# #
# # # #
# # # # # ## # # # # # ## ## #
# # # # #
# ##
#### # ##
# # # ### # ## # # # # ## # #
#
#
# # # # ## ## # ## #
# # # # ####
#
## ## ### ### # # ## ## #
# # ## # ### #####
# # #
##
#
# #
#
# ## # # # ## # ##### ## # # # # # # # # # # # # ## # #
#
## # # # #
# # ## ## ### ## ## # # # ## # # # ###
#
## ### # # #### ## ### # # # # # #
### # ### ## # # ### #
# #### ## ### ###
### # ## ## ## ## #
##
# # #
##
# ## #### #
## # ### # ##
# ## # #
# ## ### # ## ##
# ## ### #
# ## ## # # ## # ######### ## ## # # # ## # # #
### ### #
# #
# ## # ## # # # # # # ## #
# # ##
#
## # ##
#### ## ## # # ## # #
# ### ## # # # # # # #
# #
# ## ## # # # # ##
## # # # #
## # # # # # # ##### #
## #
# # #
# ##
# ######## # # #
# # # # #
# # ### ### # ## #
# ##
# # # # # ### #
# # #
## # #
# ## ### ##
## # # # #
# ## ## #
## # # # # # #
# #
# ## # # # #
# ##
# # #
# ##
# # # # #
### # #
# ##
##
0 8 16 24 Kilometers
Fig. 14: Systematic grid used to select a sample of 382 control points to determine
the relative extent of the community groupings in the Gorongosa National Park.
There are several more communities than those listed here. They can however not
easily be individually mapped either because they occur as very small units or
because they are intergraded into larger units.
Thickets on termitaria e.g. occur as small discrete patches that can not be mapped at
the scale of the GNP. However, this community represents a key resource to
browsers. Termites and termite mounds play a very important functional role in this
ecosystem.
Table 8: Plant communities identified in the different landscapes and regions of the
Gorongosa National Park. Note that several plant communities occur across different
landscapes. Communities in the Gorongosa Mountain Region are mostly according
to Muller et al. (2008).
Table 9: List of corresponding Plate numbers for the description of specific plant
communities in the Gorongosa National Park.
Plate 11: Open to Closed, Tall Hyphaene and Borassus Palm Veld
Plate 12: Colophospermum mopane Tall Closed Woodland. Total extent is 703 ha.
The diversity and spread of alien invasive species is relatively limited considering the
size and mesic character of the terrestrial systems of the Gorongosa National Park.
However, a number of waterweeds are prominent.
o This species has been planted near homesteads in the Midlands Moist
Miombo Landscape. It does appear to be relatively stable and may
only be a weak invader in that environment.
The Gorongosa environment is very conducive for the spread of these species as
well as for the establishment and spread of a number of other very invasive species.
Management must be aware of this. Immediate action and thorough follow-up are
essential if one wants to avoid massive control costs in future.
Table 10: Coordinates of alien plants observed during the current study.
6.1. Introduction
It is well documented that decades ago the Gorongosa National Park used to support
very high densities of a wide range of wildlife. Following the massive reduction in
animal numbers as a result of the civil war, a substantial recovery has been
experienced for some species. Other species such as zebra and blue wildebeest
however have been reduced to extremely low numbers that are unlikely to recover to
any significant extent without introductions from outside.
The large reduction in grazing and browsing pressure for at least three decades is
expected to have resulted in some significant changes. In particular, the grazing
succession whereby the impact of one species on the grass sward makes it more
easily accessible to another, has largely been lost because of the dramatic reduction
in grazing pressure. Can this grazing succession be re-established?
Furthermore, natural and/or man-made changes may or may not have affected the
flooding regime of the floodplain and thus its productivity and seasonal availability of
forage. The spread of alien species such as Mimosa pigra may negatively impact on
carrying capacity.
An important question for management is thus whether the current carrying capacity
and habitat suitability are adequate in terms of the generally accepted end state
(although this end state has not been explicitly defined). If the current situation is not
adequate, what should be done from a management perspective? If management-
induced changes are not feasible or would be too costly, should a different end
objective de defined?
Given the general lack of precision in determining carrying capacity and given its
fluctuation over time (because of environmental or management factors), the current
study can only attempt a first, cursory exploration of the subject.
The different parts of Gorongosa have a very different suitability for different species.
This is as a consequence of the different structure of the vegetation and different
nutrient content as determined by the underlying soils and by the moisture- and fire-
regime. The presence or absence of a clear grazing succession as mentioned above
is also important.
The results of the subjective rating of the 139 ground plots, for their suitability for
grazers and browsers on a scale of 1 to 5, are listed in Table 11 and 12.
There is a very distinct pattern in terms of the relative suitability of certain plant
communities for grazers. Densely wooded communities, particularly in the miombo of
the Midlands and the Cheringoma Plateau Region have a very low to low value.
Communities in the Rift Valley, especially grasslands, rate as good to very good. Not
surprisingly, the inverse pattern applies for browsers. The grassland communities are
not very suitable. It is only the Acacia Sparse to Open Woodland with Saline
Grassland and the Faidherbia albida Open to Closed, Tall Woodland that offer
habitat that is of equal value to the grazers and browsers.
Community
Very Very
Low Medium Good
low good
Community
Very
Very low Low Medium Good
good
As the exact size of the different communities is not known, one cannot derive a
relative suitability across the Park. Nevertheless, it is clear that the greater part of the
Rift Valley consists of communities with high suitability values for grazers (Acacia –
Combretum, Palmveld and the different grassland communities). Even in the
miombo, the Tall Closed Miombo – Acacia nigrescens woodland that adjoins the Rift
Valley offers relatively good grazing. This latter community probably represented
some of the summer dispersal range for wildlife as documented by Tinley (1977).
Gorongosa offers much less suitable habitat for browsers. It is only some of the less
extensive communities that rate highly for browsers. The current situation mirrors the
historic pattern. The ratio of bulk grazers, concentrate grazers, mixed feeders and
browsers was approximately 65%:10%:24%:1% in 1972 when wildlife populations
were still very high. This can be compared to the feeding ratio’s of
45%:20%:20%:15% of bulk grazers, concentrate grazers, mixed feeders and
browsers respectively that is proposed by Collinson & Goodman (1982) for bushveld
areas in South Africa. It is assumed that the mixed feeders (such as elephant and
impala) consume much grass in Gorongosa. Tinley (1977) lists a substantial range of
grass species utilised by elephant and impala during the dry season (his Appendix
2).
No assessment was made of the habitat suitability for individual animal species. The
overall impression is that at present the Park remains suitable for the full range of
species that were historically present. The lack of grazing succession however has
probably led to a relative lowering of the overall extent of short grass habitat.
Although short grass habitat is still present, the lower extent could place a limit on
species such as wildebeest. Insight in distribution patterns indicate that for species
such as wildebeest it is not only the structure and quality of the grazing that is
important. The ability to avoid lion predation by using short open areas is important
for this species (Fynn 2008).
Gorongosa N.P.
Riftvalley Lake Urema
Suitability for grazers
Very low
Low
Medium
Good
N
20 0 20 40 Kilometers
Gorongosa N.P.
Riftvalley Lake Urema
Suitability for browsers
Very low
Low
Medium
N
20 0 20 40 Kilometers
Fig. 15: Relative suitability of the different landscapes for grazers and browsers.
The concept of ‘carrying capacity’ is a nebulous one with many definitions, and it is
difficult to determine in heterogeneous environments experiencing variable
environmental and resource conditions (Peel et al. 1999)1. It should therefore not be
considered as a static figure but must reflect climatic conditions and influence of
management practices followed.
Within the context of the GNP, ‘ecological carrying capacity’ is loosely defined as the
population size of a species in an area as determined by the capacity of that area to
support the individuals in that population and enable them to reproduce (adapted
from Caughley 1979 and Grossman 1984). This assessment of carrying capacity
looks at ecological carrying capacity.
During the field survey of the 139 sample plots a visual estimate of carrying capacity
was made. These estimates were used to group the different communities into
‘bands’ of increasing carrying capacity (Table 13). It must be stressed that this field
survey represents only a snapshot in time in a limited number of localities. Such a
subjective estimate is influenced by the season and phenology of the vegetation. In
summer, it is likely that a more positive assessment will be made.
How does the above compare to other estimates based on general equations and
how does this compare to other case studies?
Coe et al. (1977) established that a high degree of correlation exists in natural systems
between the herbivore biomass and annual rainfall. This allows for predicting animal
biomass from simple long-term rainfall figures. The model provides a range of stocking
densities for a given long-term mean annual rainfall. The relationship between rainfall
and carrying capacity starts deteriorating above a rainfall of 800 to 900 mm, particularly
in nutrient-poor environments. Carrying capacity can actually drop in very wet areas due
to excessive leaching of nutrients from the soil and the ensuing production of a low
quality sward. This would certainly apply to the miombo on the Cheringoma plateau.
Fritz & Duncan (1994) further refined the rainfall-related carrying capacity approach
by taking into account the habitat make-up in terms of low, medium and high fertility
substrates.
Only limited recommendations on stocking rates and carrying capacity are available
for Moçambique and further research is required (Maposse et al. 2003). A predictive
model for livestock carrying capacity has been developed for Moçambique
(Timberlake & Reddy 1986). Their study lists a carrying capacity of 2.3 ha/AU for
grassy bottomlands on heavy soils in Chokwe with a annual rainfall of 656 mm.
Tinley (1977) mentions carrying capacity estimates by Myre & Antão (1972) of 15 ha
per AU for miombo, 6 – 8 ha per AU for Urochloa savanna grasslands (similar to
1
See Appendix A for a definition of a number of technical terms used throughout the report.
Combretum – Acacia veld of the present study) and 3 – 4 ha per AU for flooded
alluvial grasslands.
Table 13: Carrying capacity for wildlife of the sampled communities in Gorongosa
(visual, subjective estimate only) (communities are listed in increasing order of
carrying capacity).
Carrying capacity
Community kg km-2 ha/LSU
Tall Closed Humid Miombo Woodland 2,571 18
Short to Tall Dry Forest & Thicket 3,158 14
Low dambo Grassland 3,333 14
Miombo undifferentiated 3,333 14
Miombo hill 3,333 14
Tall Miombo Riverine Forest 3,913 12
Piliostigma thonningii – Borassus aethiopium Closed Woodland/Dry Forest 3,971 11
Combretum Short Open Woodland in Miombo 4,500 10
Dichrostachys cinerea Tall/High Closed Shrubland 4,500 10
Tall Closed Miombo – Acacia nigrescens woodland 4,909 9
Colophospermum mopane Closed Tall Woodland 5,000 9
Acacia-Combretum-Milletia Short Open Woodland 5,625 8
Termitaria 6,429 7
Dalbergia melanoxylon Low Closed Woodland 6,429 7
Acacia – Combretum Open to Close, Short to Tall Woodland 7,290 6
Acacia Sparse to Open Woodland with Saline Grassland 7,500 6
Tall Setaria Floodplain Grassland 7,941 6
Hyphaene / Borassus / Phoenix Open to Closed, Tall Palm Veld 8,504 5
Acacia xanthophloea Open to Closed, Tall Woodland 11,124 4
Reedbed 13,333 3
Pan 13,500 3
Cynodon dactylon – Digitaria swazilandensis Short Grassland 21,429 2
Faidherbia albida Open to Closed, Tall Woodland 30,000 2
Tall Echinochloa – Vetiveria Floodplain Grassland 31,765 1
A historic stocking rate for the GNP is obtained from Tinley (1977) (Table 10.3 p181).
It is however not known whether the area was at, below or above carrying capacity at
the time of the air counts. Tinley (1977) does however mention the poor physical
condition of the hippo. It would appear that Gorongosa was stocked at very high
rates. The loss of tsessebe and roan prior to the civil war may also point to problems
of competition for resources.
The results from the predictive equations, GNP counts, case studies from Parks
further a field and the current study are synthesized in Table 14. The field estimates
fit within the overall ‘envelope’ of carrying capacity derived from other sources. It
should also be noted that recent work indicates that the ‘true’ figure for carrying
capacity lies closer to the upper limit of Coe et al. (1976) rather than the average
figure (Dr M Peel, ARC, pers. comm.). All of the above would indicate that the field
estimate can indeed be used as a guideline for initial planning and stocking of the
Park.
A coarse visual representation of carrying capacity across the area is provided in Fig.
16 by allocating a relative estimate across all of the landscapes.
Table 14: Estimate of carrying capacity for the Rift Valley based on predictive models
(based on a rainfall of 840 mm) as well as recorded stocking rates from other studies.
Estimates are ranked from lowest (top) to highest (bottom).
Carrying capacity
Information source
kg km-2 ha / AU
Gorongosa N.P.
Riftvalley Lake Urema
Landscape carrying capacity for wildlife
Very low
Low
Medium
Good
Very good (Cynodon grasslands) N
20 0 20 40 Kilometers
Fig. 16: Relative estimate of carrying capacity for the different landscapes (with
additional detail for the Cynodon-Digitaria plant community).
6.4. Implications
It is pointless at this stage to try to refine the calculation of carrying capacity for the
Park. This is not only because of the shortcomings in the mapping. Carrying capacity
is not static. It depends on rainfall and is influenced by feedback from the wildlife. In
the particular case of Gorongosa, there is no fixed boundary between the tall
Echinochloa, short Cynodon and tall Setaria grasslands. These communities expand
and retract in reaction to the length and depth of inundation. The inundation pattern
will vary from year to year and includes extreme high levels such as those
experienced in early 2008.
It is clear from the range of estimates, that the individual plant communities are
capable of supporting the same order of magnitude in stocking rate as was
historically experienced in Gorongosa. Even without a precise map of the extent of
the plant communities, it is obvious that for example the Acacia-Combretum
community occurs over a very large extent (see also Table 7). This community has a
medium high carrying capacity. The key resource communities on which the wildlife
historically relied during the dry season, namely the short floodplain grasslands still
occur. Their extent however has probably been reduced through a collapse of the
grazing succession.
There are three main recommendations for the veld and wildlife management of
Gorongosa:
x The grazing succession must be re-established. This will require the presence
(and concentration) of megaherbivores such as elephant, hippo, buffalo or
white rhino. Several areas were identified where the potential exists for a
switch from tall (Setaria) grasslands to short Cynodon grasslands under the
influence of heavier grazing (Fig. 17).
Fig. 17: Ground plot 94 – south of Lake Urema on 21 September 2006. Top photo:
short Cynodon dactylon – Digitaria swazilandensis in foreground with tall Setaria
incrassata in middleground. Faidherbia albida trees in the background. Bottom
photo: Setaria is presently enchroaching on the Cynodon. However, higher grazing
pressure would reverse this trend. The Cynodon-Digitaria short grassland could
extent for 30 to 40 m back to the Faidherbia trees.
Table 15: Gaps and shortcomings in the current study and outcomes.
Fig. 18: Comparison between historic and current situation by means of photographs
(Hippo House). Top photo: 1970’s – observe tall woodland around Hippo House.
Bottom photo: 2006 – no trees.
8. CONCLUSION
The landscapes and plant communities of the Gorongosa National Park and its
Buffer Zone are very diverse and occur in a complex and intricate pattern. Although
the broad relationships between the vegetation and the environment are known,
there is insufficient quantitative information to accurately model and map the different
floodplain grasslands.
The habitat of the Park remains extremely suitable for grazers although the collapse
of the grazing succession has led to a reduction in the extent of the short Cynodon-
Dactylon grasslands which represent a key resource to the herbivores.
The carrying capacity of the Park is high and should allow stocking rates that are
comparable to historic levels.
9. ACKNOWLEDGEMENTS
Staff of the Carr Foundation and personnel from the GNP provided much logistical
support during fieldwork and staying over at Chitengo. Franziska Steinbruch provided
much GIS data including the ASTER and LISS IV satellite imagery that was used.
The provision of a helicopter (ably piloted by Tosch Ross and Bertus Reyneke)
enabled invaluable ground truthing and led to a better perspective on the GNP, its
landscapes and its functioning. Greg Carr is gratefully acknowledged for making all of
this possible.
10. REFERENCES
Acocks J.P.H. 1975. Veld types of South Africa. Memoirs of the Botanical Survey of
South Africa. No 40: 1-128.
Anderson J.L., Beilfuss R., Perreira C. & Zolho R. 2006. The proposed strategy to
reintroduce and supplement wildlife populations in Gorongosa National Park,
Moçambique. Unpublished report to the Carr Foundation.
Beilfuss R., Moore D., Bento C. & Dutton P. 2001. Patterns of vegetation change in
the Zambezi Delta, Mozambique. Program for the sustainable management of
Cahora Bassa Dam and the Lower Zambezi Valley. Working Paper 3.
Beilfuss R., Fiske G., Kerrison P., Laporte N., Lefebve P., Merry F. & Zolho R.
2007.Gorongosa Mountain Status Assessment and Opportunities for Sustainable
Agriculture Development. Unpublished report by the The Carr Foundation & The
Woods Hole Research Center, USA
Brondizio, E., E. Moran, P. Mausel, and Y. Wu. 1996. Land cover in the Amazon
estuary: Linking of the thematic mapper with botanical and historical data. Photogramm.
Eng. Remote Sens. 62:921-929.
Codd L.E.W. 1951. Trees and shrubs of the Kruger National Park. Memoirs of the
Botanical Survey of South Africa. No 26:1-192.
Coe M.J., Cumming D.H. & Phillipson J. 1976. Biomass and production of large
African herbivores in relation to rainfall and primary production. Oecologia 22:341-
354.
Coetzee B.J. 1983. Phytosociology, vegetation structure and landscapes of the Central
District, Kruger National Park, South Africa. Dissertations Botanicae. FL-9490 Vaduz,
Germany.
Collinson R.F.H. & Goodman P.S. 1982. An assessment of range condition and large
herbivore carrying capacity of the Pilanesberg Game Reserve, with guidelines and
recommendations for management. Inkwe 1:1-54.
Congalton, R.G., K. Green, and J. Teply. 1993. Mapping old growth forests on National
Forests and Park Lands in the Pacific Northwest from remotely sensed data.
Photogramm. Eng. Remote Sens. 59:529-535.
East R. 1984. Rainfall, soil nutrient status and biomass of large African savanna
mammals. Afr. J. Ecol. 22:245-270.
ESRI. 1997. Arc, Grid and ArcView Command References. Environmental Systems
Research Institute, Redlands, California, USA.
Fritz H. & Duncan P. 1994. On the carrying capacity for large ungulates of African
savanna ecosystems. Proc. R. Soc. Lond. 256: 77-82.
Fynn R.W.S. 2008. Keystone herbivores and key resource areas in Kruger National
Park: unlocking biodiversity and large herbivore abundance. Presentation at the 6th
Annual Science Network Meeting, Skukuza, Kruger National Park. 21st – 25th April
2008.
Gertenbach W.P.D. 1983. Landscapes of the Kruger National Park. Koedoe 26:9-121.
Hodgson, M.E., J.R. Jensen, H.E. Mackey, and M.C. Coulter. 1988. Monitoring wood
stork foraging habitat using remote sensing and geographic information systems.
Photogramm. Eng. Remote Sens. 54:1601-1607.
Leemans, R. & Cramer, W. 1991. The IIASA database for mean monthly values of
temperature, precipitation and cloudiness of a global terrestrial grid. International
Institute for Applied Systems Analysis (IIASA). RR-91-18.
Low A.B. & Rebelo A.G. (Eds) 1996. Vegetation of South Africa, Lesotho and
Swaziland. Dept. Environmental Affairs & Tourism, Pretoria.
Maposse I.C., Muir J. & Alage A.A. 2003. Status of range and forage research in
Mozambique. Afr. J. Range For. Sci. 20(1): 63-68.
Muller T., Wursten B., Chapano C., Ballings P., Mapuara A., Beilfuss R., Massungue
A. & Wild R. 2008. Vegetation Survey and Management Recommendation for Mount
Gorongosa. Unpublished report to the Carr Foundation
Myre M. & Antão L.R. 1972. Reconhecimento pascƭcola ao Vale de Save. I.I.A.M.
Communicações No. 75, 180 pp.
Muldavin E.H., Neville P. & Harper G. 2001. Indices of grassland biodiversity in the
Chihuahuan Desert Ecoregion derived from remote sensing. Conservation Biology
15(4):844-855.
Palgrave M.C., van Wyk A.E., Jordaan M., White J.A. & Sweet P. 2007. A
reconnaissance survey of the woody flora and vegetation of the Catapú logging
concession, Cheringoma District, Mozambique. Bothalia 37(1): 57-73.
Peel M.J.S., Biggs H. & Zacharias P.J.K. 1999. The evolving use of stocking rate
indices currently based on animal number and type in semi-arid heterogeneous
landscapes and complex land-use systems. Afr. J. Range & Forage Sci. 15(3): 117-
127.
Pienaar U. de V. 1963. The large mammals of the Kruger National Park – their
distribution and present day status. Koedoe 6: 1-37.
Solomon M., Zambatis N., Biggs H.C. & Maré N. 1999. Comparison of classifications
commonly used for templates for management, scientific and GIS work in the Kruger
National Park. Koedoe 42(2):131-142.
Stalmans M. 2006. Tinley’s plant species list for the Greater Gorongosa Ecosystem,
Moçambique. Unpublished report by International Conservation Services to the Carr
Foundation and the Ministry of Tourism.
Stalmans M. & Beilfuss R. 2007, Long-term plan for Gorongosa National Park
vegetation monitoring at multiple scales. Working draft, Carr Foundation.
Stalmans M. & Carvalho F. 2002. Plant communities and landscapes of the ‘Parque
Nacional do Limpopo’. Unpublished report to the Peace Parks Foundation.
Stalmans M. & Wishart M. 2005. Plant communities, wetlands and landscapes of the
Parque Nacional de Banhine, Moçambique. Koedoe 48(2):43-58.
Stalmans M. 2006. Tinley’s plant species list for the Greater Gorongosa Ecosystem,
Moçambique. Unpublished report by International Conservation Services to the Carr
Foundation and the Ministry of Tourism.
Timberlake J.R. & Reddy S.J. 1986. Potential pasture productivity and livestock carrying
capacity over Mozambique. Communicacão 49, Serie Terra e Agua, INIA, Maputo.
Tinley. K.L. 1977. Framework of the Gorongosa Ecosystem. Ph.d. thesis. University
of Pretoria.
Van der Schijff H.P. 1957. ‘n Ekologiese studie van die flora van die Nasionale
Krugerwildtuin. Deel 1. D.Sc. thesis, University of Potchefstroom for CHE,
Potchefstroom.
Van Wyk P. 1972. Trees of the Kruger National Park. Vols I & II. Cape Town, Purnell.
Venter F.J. 1990. A classification of land for management planning in the Kruger
National Park. Ph.D. thesis, University of South Africa, Pretoria.
Wild H. & Barbosa L.A.G. 1967. Vegetation map of the Flora zambesiaca area. Flora
zambesiaca Supplement. Collins, Salisbury, Rhodesia.
Animal Unit – a standardised unit that is used to translate domestic livestock and wild
herbivores to a common denominator. The AU used in this report denotes an average
cow of 450 kg live weight (= 1 Animal Unit or AU).
Browser – an animal that uses browse which is the proportion of the woody
vegetation that is available for consumption by animals.
Sour (sour veld) – veld in which the forage plants become unacceptable to the
animals and less nutritious upon attaining maturity. There is a sharp drop in protein
content in late-summer/autumn. Although a large amount of forage might remain
standing throughout the winter (dry season), it is of such low quality that it cannot fulfil
the nutritional requirements of most herbivore species.
Sweet (sweet veld) – veld in which the forage plants retain their acceptability and
nutritive value after maturity or in which different plants are acceptable at different
times so that the veld can be utilized by animals at all times of the year.
Stocking rate – the area of land in the system of management that is allocated to
each animal unit in the system (ha/AU or AU/ha).
Adapted from:
Gabriel H.W. & Talbot S.S. 1984. Glossary of landscape and vegetation
ecology for Alaska. BLM–Alaska Technical Report 10. U.S. Department
of the Interior, Anchorage, Alaska.
gorongosa_03_05sep05
serra_1_06sep05
gorongosa_06_05sep05
serra_2_06sep05
png_02_05sep05
png_03_25sep05
gorongosa_10_06sep05 png_06_05sep05
png_08_01sep05
png_07_25sep05
png_10_05sep05
&
png_12_01sep05
png_13_05sep05 png_15_01sep05
& Chitengo
Lake Urema
Gorongosa NP
Bufferzone
LISS IV September 2005
N
20 0 20 40 60 Kilometers
gorongosa_02_13oct05
gorongosa_05_13oct05
&
& Chitengo
Lake Urema
Gorongosa NP
Bufferzone
LISS IV October 2005 N
20 0 20 40 60 Kilometers
gorongosa_08_20sep06
png_04_20sep06
gorongosa_09_25sep06
gorongosa_11_25sep06
&
gorongosa_12_25sep0
gorongosa_04_02nov06
serra_3_01nov06
gorongosa_07_02nov06
png_01_01nov06
png_05_01nov06
png_09_01nov06
&
png_11_02nov06
gorongosa_13_01nov06
png_14_02nov06
& Chitengo
Lake Urema
Gorongosa NP
Bufferzone
LISS IV November 2006 N
20 0 20 40 60 Kilometers
APPENDIX C: Plant species recorded during this study that are not
contained in the Tinley (1977) list.
Woody classes
Woody density
Grasslayer density
Woody height
Grasslayer height