Professional Documents
Culture Documents
MBOMIPA PROJECT
Matumizi Bora ya Malihai Idodi na Pawaga
Sustainable Use of Wild Resources in Idodi and Pawaga
November 2002
by
ECOSYTEMS CONSULTANTS
London and Nairobi
MBOMIPA Project
Iringa District Natural Resources Office
P.O.Box 398
IRINGA
3 Results .....................................................................................................................................7
3.1 General observations.......................................................................................................7
3.2 sample survey ..................................................................................................................7
3.2.1 Species of principal interest ....................................................................................7
3.2.2 Other species..........................................................................................................7
3.2.3 Surface Water .........................................................................................................8
3.3 Total Count of Buffalo.......................................................................................................8
4 Results of all sample surveys ............................................................................................ 12
4.1 Comparative analyses................................................................................................... 12
4.2 Discussion..................................................................................................................... 12
5 Hunting quotas..................................................................................................................... 14
5.1 Buffalo ........................................................................................................................... 14
5.2 Other species ................................................................................................................ 16
6 Final conclusions and recommendations ........................................................................ 18
6.1 Diversification of quota.................................................................................................. 18
6.2 Quota setting procedures.............................................................................................. 18
6.3 Ground monitoring ........................................................................................................ 18
6.4 Aerial monitoring ........................................................................................................... 19
7 Acknowledgements ............................................................................................................. 19
8 References............................................................................................................................ 20
Wildlife is a critical resource for stakeholders in the MBOMIPA Project area (see Figure 1). At
present project villages derive a variety of financial and other benefits all of which depend on
the abundance of wildlife – especially large mammal and game1 populations - and its
sustainable utilisation, both consumptive and non-consumptive. In order to facilitate such
utilisation and its future development, project implementers and stakeholders need adequate
and up-to-date information about the distribution and status of key wildlife species in the project
and adjacent areas.
The present volume reports and discusses the results of the sixth in the proposed series of
aerial surveys. This took place towards the end of the 2002 dry season, between 14 and 25
October. The work consisted of a sample survey throughout the project area (16-22 October),
followed by a total count of buffalo in the southwestern sector of the Lunda Mkwambi Game
Controlled Area (23-24 October).
The survey took place after an interval of 18 months since the previous survey, because of
uncertainty surrounding the future of the Project and arrangements for management of its 14-
month extension (November 2001 to December 2002). Once resolved, a decision was reached
to make the survey reported here the last. Consequently this report includes a comparative
analysis of the entire series of six surveys, and as such serves also as a Final Report. It should
be read in conjunction with its five predecessors (Ecosystems 1999b, c; 2000a, c; and 2001b).
In 1994 and 1995, the Ruaha Ecosystem Wildlife Management Project (REWMP) flew three
wildlife surveys in a 6,250 km2 survey area comprising the Rift Valley section of Ruaha National
Park (RNP), and the adjacent parts of Lunda Mkwambi Game Controlled Area South (LMGCA-
South), according to a method called the Systematic Reconnaissance Flight (usually known by
its acronym, SRF), details of which may be found in Norton-Griffiths (1978).
Consequently the MBOMIPA project has an established baseline against which new data may
be compared in order to assess medium term trends in game status in the project area,
provided the same SRF method is used at the same sample intensity, following if possible, the
same flight transects. The nominal sampling parameters of the present survey (see Box
entitled “Survey Configuration”), therefore matched those of its predecessors as closely as
possible. The transect coordinates used during the REWMP surveys were found and mapped,
meaning it has been possible to follow the same flight lines in all subsequent surveys using a
Garmin 195 GPS navigation computer (Figures 2 and 3).
1
A term referring to those species commonly hunted under license for meat or trophies.
MBOMIPA Project Area
(
LMGCA North
Rungwa Game Reserve
Kisanga
Kinyika Mboliboli
Kimande
Mbuyuni Isele Itunundu
Mkombilenga
Ilolo
Mafuluto Magozi
Ruaha National Park Malinzanga Luganga
er
CA
iv
R
G
ha
LM
ua
R
at
re Nyamahana
Ma
G
po
Tu
go eng
ng
Makifu Idodi
am
ro
al
Mkupule
a
Usangu Game Reserve
Mahuninga
0 20 40
Kilometers
Rivers RNP/LMGCA boundaries Village boundaries Mtera Dam LMGCA Lunda-Mkwambi Game Controlled Area
Sixth Aerial Game Survey 3
FIGURE 2
9,250,000
9,240,000
9,230,000
9,220,000
9,210,000
9,200,000
9,190,000
9,180,000
9,170,000
9,160,000
9,150,000
9,140,000
9,130,000
9,120,000
9,110,000
9,100,000
9,090,000
590,000
600,000
610,000
620,000
630,000
640,000
650,000
660,000
670,000
680,000
690,000
700,000
710,000
720,000
730,000
740,000
750,000
760,000
770,000
780,000
790,000
800,000
810,000
820,000
830,000
840,000
850,000
860,000
4 MBOMIPA Game Surveys and Participatory Monitoring Programme
FIGURE 3
0 20 40
Kilometers
Sixth Aerial Game Survey 5
The design for MBOMIPA’s aerial monitoring programme includes the entire REWMP survey
area, plus a new area which was added to enable coverage of the more northerly and
previously excluded parts of the Lunda Mkwambi GCA. These are referred to respectively as
the “Old” and “New” survey areas. For analytical purposes, the Old area is further broken down
into those parts within the Ruaha National Park, and those parts in the LMGCA2 (see Table 2.1
and Figure 4). All areas, old and new, amount to 8,325 km2.
In common with all MBOMIPA surveys to date (but not the REWMP series), the in-flight
records for cattle, sheep and goats (shoats) and some wildlife have been corrected from
oblique photos taken of all groups consisting of more than ten animals3. In the third survey and
subsequently, counts have been made of bomas in use, recently abandoned bomas, and
boma scars. These data are collected to help determine whether any changes in livestock
numbers noted are due to sampling error, or to movements in or out of the study area.
Data Manager:
Ms C. Geddes-Mathews (Ecosystems)
2
Between the fifth survey in April 2001 and this one in October 2002, a dispute between Iringa and Mbarali Districts regarding the
Mkupule area was settled with both parties agreeing that the most southwesterly part, formerly known as “Mkupule Wilaya” is in fact
in the Usangu Game Reserve which is in Mbarali District. The maps generated for this report reflect this boundary change. However,
the total area surveyed remained exactly the same, and in order to enable robust comparisons for the “GCA stratum” between this
survey and the others, data for the “lost” portion are still treated as within the GCA for analytical purposes.
3
Taken on 400ASA colour slide film, for later interpretation under a binocular microscope.
6 MBOMIPA Game Surveys and Participatory Monitoring Programme
FIGURE 4
As is typical for the late dry season, tree foliage was generally absent, and all grass either burnt
or reduced to a trampled straw litter. Despite this animal condition appeared to be relatively
good, and the availability of surface water was greater than that recorded in a previous dry
season (Ecosystems 1999c). Several relatively minor tributaries of the Great Ruaha still held
pools (e.g. the Mdonya), the Little Ruaha was in full flow, and numerous springs were active in
the Mkupule area. Similarly, the level of the Mtera dam was higher than had been observed in
previous dry seasons. All this is attributed at least in part to the two consecutive wet seasons of
2001 and 2002, which apparently were both wetter than usual (for a description of conditions
during the former, see Ecosystems 2001b).
As in previous dry season surveys, the daily flights were commenced about an hour earlier
than in the wet season (at 0700h as opposed to 0800h).
!"!
The wildlife species of principal interest to project implementers and stakeholders are those
larger mammals that either can be hunted on license, or which form the basis to tourism, or
both. These include buffalo, eland, elephant, giraffe, hartebeest, impala, kudu4, reedbuck,
sable, warthog, waterbuck and zebra. Also of interest are the livestock which compete with
wildlife for habitat, food and water, namely cattle, sheep, goats and donkeys.
The distribution maps and population estimates for these species resulting from this latest dry
season survey are presented in Appendix A. For each species, the map of the latest
distribution is followed by all the distributions recorded so far shown on a single sheet for ease
of comparison. In the Old area, the recent estimates (for all six MBOMIPA surveys to date) are
listed with data collected under REWMP (Ecosystems 1999b, c; 2000a, c; and 2001b; and
REWMP 1994a,b and 1995).
#
It is normal practice during an SRF survey to record all species actually seen, irrespective of
whether they are of principal interest. Owing to their relative rarity, small size, cryptic nature or
all three however, the estimates of many of the species seen return very wide confidence limits
4
Both Lesser and Greater Kudu occur in the project area, but it is not possible to distinguish between these during an aerial
survey.
8 MBOMIPA Game Surveys and Participatory Monitoring Programme
and so normally are not reported. For the record however, Table 3.1 summarises the numbers
of these other species counted in the sampling strip by area5.
As shown in Table 3.2, the records of elephant bones seen provide an interesting indicator of
the contrasting visibility conditions prevailing in the different seasons. No fresh elephant
carcasses were seen in any of the surveys.
Table 3.2: Elephant bones observed in the samples, by season and year
WET DRY
1999 7 30
2000 5 40
2001/2 3 21
"$"
%
&&&
As can be seen from Table 3.3 and Figure 6, the majority (86%) of the buffalo located during
the total count was in the National Park. The remainder was seen in the area previously known
as “Mkupule Wilaya”, but which is now acknowledged to be in Mbarali District and therefore
part of the Usangu Game Reserve and not part of Lunda Mkwambi. No buffalo were seen in
the LMGCA proper.
AREA COUNT
National Park – North bank 5
National Park – South bank 631
Total RNP 636
Usangu Game Reserve 105
Total UGR 105
MBOMIPA village areas 0
Total LMGCA(S) 0
5
These are untransformed count data from both observers combined.
6
Early rain pre-empted a description of surface water distribution for the 2000 dry season (Ecosystems 2000c)
Sixth Aerial Game Survey 9
FIGURE 5a
Water present
0 20 40
Kilometers
10 MBOMIPA Game Surveys and Participatory Monitoring Programme
FIGURE 5b
Dry Season (October) 1999 Dry Season (November) 2000 Dry Season (October) 2002
Sixth Aerial Game Survey 11
FIGURE 6
Total Count
1 - 50
0 10 20
51 - 150
Kilometers
151 - 250
> 250
As is the nature of the method, all the estimates obtained in the present series of six surveys
demonstrate considerable variation, both within each census (in terms of confidence limits) and
between censuses (marked variation within a set of population estimates for an individual
species). However, to all intents and purposes, the operational procedures adopted for each of
the six censuses were strictly comparable7. As intended, this validates certain statistical tests to
gauge the extent to which these two sources of variation could be ascribed to random
(sampling) error rather than to seasonal effects, or even to time trends. If sampling error can be
shown to explain most of the variation noted, this in turn validates a procedure in which the
separate estimates in a series are pooled to arrive at a single “merged” estimate with low
variance.
An independent analysis of the full data set in respect of species of principal interest was
carried out to this end (Appendix B). This focussed on the 6,250km2 “Old (All)” stratum (see
Table 2.1 and Figure 4), because this encompasses both the consumptive use zone (the GCA)
and its resource buffer (the RNP).
The conclusion reached is that the best model from a practical viewpoint is to treat the six
estimates for each species as independent samples from the same population, and pool the
results to obtain a more precise, single estimate. The procedure for merging estimates, which
involves weighting each individual result in inverse proportion to its variance, is detailed in
Norton-Griffiths (1978). The merged estimates and associated confidence limits so derived are
included in the tables of Appendix A.
!"
The seasonally consistent dry-to-wet increases noted for cattle and zebra in 1999 and 2000
have been upheld only for zebra (Table 4.1).
Table 4.1: Estimated numbers of cattle and zebra, by year and season
CATTLE ZEBRA
Wet Dry Wet Dry
1999 64,171 36,724 5,840 3,176
2000 56,633 26,756 6,304 3,100
2001/2 60,093 72,647 5,090 2,012
The latest cattle estimate, which is the largest in the series, effectively ends speculation that
cattle might move out of the study area during the dry season and back in the wet. Had this
been a real phenomenon, one would have expected the data for sheep and goats, donkeys,
bomas in use, and recently abandoned bomas to show a consistent pattern when in fact this
was never the case (Figure 7).
7
In terms of height, speed, strip width, observers and photo-interpreters etc
Sixth Aerial Game Survey 13
Livestock by season
80000
60000
40000
20000
0
Wet 99 Dry 99 Wet 00 Dry 00 Wet 01 Dry 02
Except for zebra, none of the seasonal differences summarised in Figure 8 are statistically
significant (Appendix B). Only two non-random explanations for the exception are possible.
Either zebra really are moving in and out, or some season-sensitive attribute of the
methodology is having an effect. As noted previously, there is no convincing evidence for the
first alternative (Ecosystems 2001b). However, the theory that zebra are simply harder to spot
and count in dry conditions is now even more plausible. Subjectively, the observers feel that
the extreme background contrast of the wet season makes zebra easier to spot, while in the
dry season they are much better camouflaged.
8000
6000
4000
2000
0
Wet 99 Dry 99 Wet 00 Dry 00 Wet 01 Dry 02
Although seasonal effects are effectively absent, with wildlife overall there does appear to be a
weak but consistent time effect, with populations showing a general decrease over the 43
months of the survey programme (Appendix B). It is not easy to determine which species are
the most significant contributors to this overall trend but warthog is likely to be one, having
returned unusually low estimates in the sixth survey for which there is no ready explanation.
The general decrease is not particularly strong, but must be borne in mind when considering
management options (6.1.4).
The only species for which the data indicate a convincing positive trend over the study period,
is the elephant (Appendix B). The distributions recorded in all six surveys are remarkably
similar and, although there is significant usage of areas outside the National Park, the bulk of
the elephant population appears to be resident within the Old (All) stratum.
14 MBOMIPA Game Surveys and Participatory Monitoring Programme
#$%&%'(
The consultants have been asked to review the quota recommendations made in their
previous reports in the light of the latest results and analyses, particularly for buffalo, with a
view to applying any revisions recommended to the 2003 hunting season.
#)"**
The SRF and total count data provide convincing evidence that buffalo use the GCA more in
the dry season than the wet (Figure 9), although some of the difference could be due to relative
“countability”. Either way, this does not invalidate the conclusion reached in previous reports
that there are few buffalo, if any, permanently resident in the LMGCA(S). As noted previously,
this has certain important implications regarding a sustainable offtake strategy for the species
in that area.
Even though it could only be exploited inside the GCA, one approach could be to treat the
population of the entire Old (All) stratum8 as the basic resource. Whilst in theory this might be
sustainable, such an approach is unlikely to find favour either with the principal proprietors of
the resource, TANAPA, or the Director of Wildlife. A less contentious approach would be to
base the quota on the data now available for the GCA itself which, for buffalo only, exists in two
sets.
The first consists of data from the total counts. Because buffalo hardly use the GCA in the wet
season (or are very hard to see there at that time), and because the wet season anyway does
not coincide with the hunting season, the wet season data are neither robust nor relevant and
can be ignored. The average for all three dry season total counts is 142 (see bold figures in
Table 5.1).
The second data set comes from the SRF surveys which yield a merged estimate for the GCA
of only 45 animals, but the 95% confidence limits are very wide, at plus or minus 250%. The
true number is likely to be closer to the upper confidence limit than the lower, as the latter is
unbelievable: even a simple average of the actual numbers seen in the sampling strip is
greater than 45, wet seasons included (Table 5.1). The upper limit of 157 animals however is
believable, because it is consistent with the dry season average for all total counts of 142.
8
which includes the National Park’s entire “Rift Valley” sector
Sixth Aerial Game Survey 15
FIGURE 9
MBOMIPA
Total Count of Buffalo
Wet Season (April) 1999 Wet Season (May) 2000 Wet Season (April) 2001
Dry Season (October) 1999 Dry Season (November) 2000 Dry Season (October) 2002
16 MBOMIPA Game Surveys and Participatory Monitoring Programme
The two data sets indicate that on any particular day in the dry season one may reasonably
expect there to be roughly 150 buffalo in the GCA. However, we know that on occasion
numbers can reach at least 361 (Table 5.1). This is because the buffalo population is not
resident and animals move in and out all the time.
The question is whether to base the quota on a hypothetical “average” or an actual “highest
recorded so far” number. At 9% per annum (Martin and Thomas 1991), the former would give
a quota of 14 and the latter 32. Two arguments favour the higher quota.
Firstly, the average number of buffalo actually killed in 2000 and 2001 was 14, despite
circumstances that were considered unhelpful to hunting success (Ngomello, personal
communication). This suggests a ceiling of 14 would be excessively conservative. However the
most compelling argument relates to the open nature of the system, meaning that even if every
single buffalo were shot they would soon be replaced through immigration from the Park.
Consequently, there can be little doubt that the higher quota would be sustainable.
However, now that its true boundary with the Usangu Game Reserve has been recognised,
the GCA has in effect lost the greater part of what hitherto had been its best buffalo area.
Whether such a quota is realisable within what remains is another matter, as this will depend
on actual occupance levels there during the hunting season. Since the hunting and dry
seasons overlap, the prospects are good, but only the ease or difficulty actually experienced in
trying to fulfil any quota will indicate its suitability within the concept of the GCA being an
independently managed buffalo range.
It is recommended therefore that in 2003 the overall buffalo quota should remain at 32.
Since there are no total count data for the other species on quota, the only objective basis on
which to revise quotas is to use the merged estimates now available. However, the buffalo
analysis above strongly suggests that the upper confidence limits should be used for the
following reasons. Firstly the total count data for buffalo provide a cross-check that indicates
these to be the more realistic, and secondly because the argument that the open nature of the
system provides a fail-safe buffer is applicable to all species not just buffalo. At the same time it
should be noted that the SRF method is known to yield conservative estimates, and that for
several of the species concerned an under-estimate is anyway unavoidable either due to body
size (dik dik), cryptic colouration (kudu), nocturnal behaviour (bushpig), or small populations
with clumped distributions (sable, eland, waterbuck).
Based on assumed populations calculated where possible from the upper confidence limits of
the merged estimates, revised quotas can be derived as shown in Table 5.2:
Sixth Aerial Game Survey 17
Species AssumedMaximum Theoretically Trend apparent from Quota Quota Quota ceiling
population in
sustainable sustainable aerial sample data awarded awarded recommended
LMGCA South
offtake as quota in 2001 in 2002 for 2003
% (Note 1)
SPECIES AVAILABLE TO RESIDENT HUNTERS
Bushpig ? 20 ? Probably stable 20 20 (20) (Note 2)
Dik Dik ? 15 ? Probably stable 40 40 40(Note 3)
Hartebeest ? 10 ? Down 0 0 0 (Note 4)
Impala 652 20 130 Stable 130 130 130
Warthog 111 20 22 Stable/down 30 30 20
Eland 190 10 19 Stable 5 5 10
NOTES:
Although the Director of Wildlife refused to renew the special dispensation granted in 2001 to
hunt zebra, kudu, sable and waterbuck (species usually reserved for “tourist” hunters)9, there
are no biological grounds for this decision. The application therefore must be repeated if
revenues from the GCA, and for the communities, are to be maximised. Pressure must be
maintained for more traditional “tourist” species to be made available within CBNRM
programmes if the latter are to succeed. Ecologically, there are no grounds for refusing the
very conservative quotas recommended above. Zebra are abundant, and the other species
proposed are definitely widespread and, being hard to spot, certainly more numerous than the
SRF data indicate. The available ground monitoring data support this assertion (Ecosystems
2001a).
!"
As emphasised and explained elsewhere (Ecosystems 2000b and 2001a), a participatory
approach to quota setting must be adopted that “triangulates” all available relevant data. In this
way the implications of each data set can be reconciled, and a consensus reached among all
participants as to a suitable adjustment, if any, to the previous year’s quota for each species.
Despite two key officers having visited Zimbabwe to gain first hand experience of this process,
the concerned district and wildlife authorities have yet to inaugurate a participatory approach to
quota setting.
The primary data sets that will support this approach to quota setting are those provided by
aerial and ground monitoring programmes. Conclusions and recommendations regarding each
are given below.
#!$!
A dedicated consultancy to design and introduce a participatory ground monitoring system in
the study area was carried out in July 2000 (Ecosystems 2000b). Monitoring by Village Game
Scouts began soon after, but the standard of data collection was variable, and by March 2001
no data had yet been entered on computer. To ensure that the relevant baseline indices would
be set by April 2001, so that one year later it would be possible to derive trends that can inform
the quota setting process, a follow up consultancy was undertaken and certain revised
procedures recommended (Ecosystems 2001a).
As the previous aerial survey report in this series showed, even the limited ground data then
available proved useful to the quota setting process, which hitherto had relied exclusively on
aerial data (Ecosystems 2001b). Unfortunately data entry problems have continued to affect
the ground monitoring programme, and at the time of writing there is no new analysis of those
data available to inform the recommendations made above. Despite this, the potential value of
the ground monitoring programme remains great, and if possible it should be carried forward
indefinitely. A further “follow-up” consultancy is planned to try and put the programme into a
position that will facilitate this.
9
and in 2000, for zebra only
Sixth Aerial Game Survey 19
%&$!
The data analyses in this and the other reports in the series have proved the utility of
professionally conducted aerial surveys for assessing resource availability and sustainable
offtake levels in a managed hunting area. Some might argue that aerial surveys are a luxury
that few equivalent programmes will be able to afford, whereas in fact they simply must be
done if CBNRM programmes involving consumptive utilisation are to be managed objectively
and sustainably. This is because the ground methods that could return quantitative data of
comparable quality are technically and logistically prohibitive. The more “user-friendly”
approach recommended and discussed above can track trends, but cannot give any insight
into actual numbers of animals.
Aerial surveys are therefore essential for “calibrating” any ground-based effort in quantitative
terms. However, one important fact to be noted from the MBOMIPA experience is that even
with a 10% sample, the confidence limits for most of the relevant species are very wide. As a
result, the method proved too insensitive to pick up anything but “noise” when repeated at the
frequency applied in this series of surveys (Appendix B).
This suggests that an interval between surveys greater than the duration of the MBOMIPA
series (43 months) would be acceptable as long as the system being monitored remains
relatively stable. Clearly however, follow-ups cannot be too widely spaced because should a
significant upward or downward trend emerge, only knowledge of the fact would allow both the
ground monitoring effort and the quota setting process to be re-calibrated. The indications of an
incipient downward trend for wildlife overall (Appendix B), underlines the need for further aerial
surveys in the MBOMIPA area. Obviously, it is important to keep the population sizes assumed
for quota setting purposes as close to reality as possible, otherwise sustainability will be
compromised.
10
The approach referred to is discussed more fully in Ecosystems 2000b and 2001a.
20 MBOMIPA Game Surveys and Participatory Monitoring Programme
+,
Ecosystems Consultants (1999a) Game Surveys and Participatory Monitoring Programme: A proposal
for technical assistance. MBOMIPA Report No. MCR2, Iringa
Ecosystems Consultants (1999b) Game surveys of Lunda Mkwambi Game Controlled Area and
adjacent areas of Ruaha National Park: First Survey, 21-27 April 1999. MBOMIPA Report No. MCR5,
Iringa
Ecosystems Consultants (1999c) Game surveys of Lunda Mkwambi Game Controlled Area and adjacent
areas of Ruaha National Park: Second Survey, 17-29 October 1999. MBOMIPA Report No. MCR7,
Iringa
Ecosystems Consultants (2000a) Game surveys of Lunda Mkwambi Game Controlled Area and
adjacent areas of Ruaha National Park: Third Survey, 29 April - 9 May 2000. MBOMIPA Report No.
MCR8, Iringa
Ecosystems Consultants (2000b) Participatory Monitoring of Wildlife Resources for MBOMIPA Project
Villages. MBOMIPA Report No. MCR9, Iringa
Ecosystems Consultants (2000c) Game surveys of Lunda Mkwambi Game Controlled Area and adjacent
areas of Ruaha National Park: Fourth Survey, 13 – 23 November 2000. MBOMIPA Report No. MCR10,
Iringa
Ecosystems Consultants (2001a) Participatory Monitoring of Wildlife Resources for MBOMIPA Project
Villages: Data Analysis and Performance Review. MBOMIPA Report No. MCR11, Iringa
Ecosystems Consultants (2001b) Game surveys of Lunda Mkwambi Game Controlled Area and
adjacent areas of Ruaha National Park: Fifth Survey, 9 – 19 April 2001. MBOMIPA Report No. MCR10,
Iringa
Martin, R. and S. Thomas (1991) Quotas for sustainable wildlife utilisation in communal lands: manual for
district councils with appropriate authority. The Zimbabwe Trust and Department of National Parks and
Wildlife Management, Harare.
REWMP (1994a) Aerial Wildlife Survey in Rift Valley sectors of Ruaha National Park and Lunda
Mkwambi Game Controlled Area: 13-19 July 1994. Ruaha Ecosystem Wildlife Management Project.
REWMP (1994b) Aerial Wildlife Survey in Rift Valley sectors of Ruaha National Park and Lunda
Mkwambi Game Controlled Area: 8-14 November 1994. Ruaha Ecosystem Wildlife Management
Project.
REWMP (1995) Aerial Wildlife Survey in Rift Valley sectors of Ruaha National Park and Lunda Mkwambi
Game Controlled Area: 1-6 June 1995. Ruaha Ecosystem Wildlife Management Project.
Taylor, R.D. (1995) Consultancy to assess the potential for community management of wildlife resources
in the Lunda-Mkwambi Game Controlled Area adjacent to the south eastern border of the Ruaha
National Park. Report to the British Development Division in Eastern Africa.
Sixth Aerial Game Survey 21
--." -
/
-,
- ,-
-
BOMAS in use
Numbers 95% C.L.
Old Area (All)
REWMP 1 - Jul 94
REWMP 2 - Nov 94
REWMP 3 - Jun 95
MBOMIPA 1 - Apr 99
MBOMIPA 2 - Oct 99
MBOMIPA3 - May00 1,072 44
MBOMIPA4 - Nov 00 881 47
MBOMIPA5 - Apr 01 730 42
MBOMIPA6 – Oct 02 874 41
REWMP 3 - Jun 95
MBOMIPA 1 - Apr 99
MBOMIPA 2 - Oct 99
MBOMIPA3 - May00 48 157
MBOMIPA4 - Nov 00 0
MBOMIPA5 - Apr 01 9 189
MBOMIPA6 – Oct 02 0
REWMP 3 - Jun 95
MBOMIPA 1 - Apr 99
MBOMIPA 2 - Oct 99
MBOMIPA3 - May00 1,023 37
MBOMIPA4 - Nov 00 883 40
MBOMIPA5 - Apr 01 729 35
MBOMIPA6 – Oct 02 874 35
New Area
MBOMIPA 1 - Apr 99
MBOMIPA 2 - Oct 99
MBOMIPA3 - May00 386 53
MBOMIPA4 - Nov 00 283 35
MBOMIPA5 - Apr 01 172 51
MBOMIPA6 – Oct 02 437 45
Sixth Aerial Game Survey 23
Sample Count
1-5
6 - 15
16 - 25
0 20 40
Kilometers > 25
24 MBOMIPA Game Surveys and Participatory Monitoring Programme
Wet Season (April) 1999 Wet Season (May) 2000 Wet Season (April) 2001
Dry Season (October) 1999 Dry Season (November) 2000 Dry Season (October) 2002
Sixth Aerial Game Survey 25
REWMP 1 - Jul 94
REWMP 2 - Nov 94
REWMP 3 - Jun 95
MBOMIPA 1 - Apr 99
MBOMIPA 2 - Oct 99
MBOMIPA3 - May00 1,245 44
MBOMIPA4 - Nov 00 922 40
MBOMIPA5 - Apr 01 333 50
MBOMIPA6 – Oct 02 528 48
REWMP 3 - Jun 95
MBOMIPA 1 - Apr 99
MBOMIPA 2 - Oct 99
MBOMIPA3 - May00 0
MBOMIPA4 - Nov 00 30 142
MBOMIPA5 - Apr 01 64 136
MBOMIPA6 – Oct 02 0
REWMP 3 - Jun 95
MBOMIPA 1 - Apr 99
MBOMIPA 2 - Oct 99
MBOMIPA3 - May00 1,245 37
MBOMIPA4 - Nov 00 893 32
MBOMIPA5 - Apr 01 271 42
MBOMIPA6 – Oct 02 528 43
New Area
MBOMIPA 1 - Apr 99
MBOMIPA 2 - Oct 99
MBOMIPA3 - May00 280 45
MBOMIPA4 - Nov 00 162 63
MBOMIPA5 - Apr 01 124 57
MBOMIPA6 – Oct 02 305 58
26 MBOMIPA Game Surveys and Participatory Monitoring Programme
Sample Count
1-5
6 - 15
16 - 25
0 20 40
> 25
Kilometers
Sixth Aerial Game Survey 27
No Window
Dry Season (October) 1999 Dry Season (November) 2000 Dry Season (October) 2002
28 MBOMIPA Game Surveys and Participatory Monitoring Programme
BUFFALO
Numbers 95% C.L.
Old Area (All)
REWMP 3 - Jun 95 63
MBOMIPA 1 - Apr 99 80 332
MBOMIPA 2 - Oct 99 1,865 249
MBOMIPA3 - May00 39 189
MBOMIPA4 - Nov 00 467 189
MBOMIPA5 - Apr 01 0
MBOMIPA6 – Oct 02 0
Merged estimate 45 250
New Area
MBOMIPA 1 - Apr 99 0
MBOMIPA 2 - Oct 99 31 181
MBOMIPA3 - May00 0
MBOMIPA4 - Nov 00 101 189
MBOMIPA5 - Apr 01 0
MBOMIPA6 – Oct 02 0
Sixth Aerial Game Survey 29
Total Count
1 - 50
0 10 20
51 - 150
Kilometers
151 - 250
> 250
MBOMIPA
Total Count of Buffalo
Wet Season (April) 1999 Wet Season (May) 2000 Wet Season (April) 2001
Dry Season (October) 1999 Dry Season (November) 2000 Dry Season (October) 2002
Sixth Aerial Game Survey 31
CATTLE
Numbers 95% C.L.
Old Area (All)
New Area
Sample Count
1-5
6 - 15
16 - 25
0 20 40
Kilometers > 25
Sixth Aerial Game Survey 33
Dry Season (October) 1999 Dry Season (November) 2000 Dry Season (October) 2002
34 MBOMIPA Game Surveys and Participatory Monitoring Programme
DONKEY
Numbers 95% C.L.
Old Area (All)
REWMP 3 - Jun 95 0
MBOMIPA 1 - Apr 99 0
MBOMIPA 2 - Oct 99 0
MBOMIPA3 - May00 0
MBOMIPA4 - Nov 00 0
MBOMIPA5 - Apr 01 91 189
MBOMIPA6 – Oct 02 0
REWMP 3 - Jun 95 83
MBOMIPA 1 - Apr 99 711 92
MBOMIPA 2 - Oct 99 315 132
MBOMIPA3 - May00 319 110
MBOMIPA4 - Nov 00 507 64
MBOMIPA5 - Apr 01 1,402 45
MBOMIPA6 – Oct 02 935 69
New Area
Sample Count
1-5
6 - 15
16 - 25
0 20 40
Kilometers > 25
36 MBOMIPA Game Surveys and Participatory Monitoring Programme
Dry Season (October) 1999 Dry Season (November) 2000 Dry Season (October) 2002
Sixth Aerial Game Survey 37
ELAND
Numbers 95% C.L.
Old Area (All)
REWMP 3 - Jun 95 58
MBOMIPA 1 - Apr 99 51 276
MBOMIPA 2 - Oct 99 47 362
MBOMIPA3 - May00 0
MBOMIPA4 - Nov 00 183 127
MBOMIPA5 - Apr 01 150 180
MBOMIPA6 – Oct 02 142 129
Merged estimate 86 121
New Area
MBOMIPA 1 - Apr 99 0
MBOMIPA 2 - Oct 99 0
MBOMIPA3 - May00 116 189
MBOMIPA4 - Nov 00 0
MBOMIPA5 - Apr 01 400 126
MBOMIPA6 – Oct 02 0
38 MBOMIPA Game Surveys and Participatory Monitoring Programme
Sample Count
1-5
6 - 15
16 - 25
0 20 40
Kilometers > 25
Sixth Aerial Game Survey 39
Dry Season (October) 1999 Dry Season (November) 2000 Dry Season (October) 2002
40 MBOMIPA Game Surveys and Participatory Monitoring Programme
ELEPHANT
Numbers 95% C.L.
Old Area (All)
New Area
Sample Count
1-5
6 - 15
16 - 25
0 20 40
Kilometers > 25
42 MBOMIPA Game Surveys and Participatory Monitoring Programme
Dry Season (October) 1999 Dry Season (November) 2000 Dry Season (October) 2002
Sixth Aerial Game Survey 43
GIRAFFE
Numbers 95% C.L.
Old Area (All)
New Area
Sample Count
1-5
6 - 15
16 - 25
0 20 40
Kilometers > 25
Sixth Aerial Game Survey 45
Dry Season (October) 1999 Dry Season (November) 2000 Dry Season (October) 2002
46 MBOMIPA Game Surveys and Participatory Monitoring Programme
HARTEBEEST
Numbers 95% C.L.
Old Area (All)
REWMP 3 - Jun 95 47
MBOMIPA 1 - Apr 99 11 188
MBOMIPA 2 - Oct 99 0
MBOMIPA3 - May00 0
MBOMIPA4 - Nov 00 142 188
MBOMIPA5 - Apr 01 0
MBOMIPA6 – Oct 02 0
REWMP 3 - Jun 95 0
MBOMIPA 1 - Apr 99 7 366
MBOMIPA 2 - Oct 99 0
MBOMIPA3 - May00 0
MBOMIPA4 - Nov 00 0
MBOMIPA5 - Apr 01 0
MBOMIPA6 – Oct 02 0
New Area
MBOMIPA 1 - Apr 99 0
MBOMIPA 2 - Oct 99 0
MBOMIPA3 - May00 0
MBOMIPA4 - Nov 00 0
MBOMIPA5 - Apr 01 0
MBOMIPA6 – Oct 02 0
Sixth Aerial Game Survey 47
IMPALA
Numbers 95% C.L.
Old Area (All)
New Area
Sample Count
1-5
6 - 15
16 - 25
0 20 40
Kilometers > 25
Sixth Aerial Game Survey 49
Dry Season (October) 1999 Dry Season (November) 2000 Dry Season (October) 2002
50 MBOMIPA Game Surveys and Participatory Monitoring Programme
KUDU
Numbers 95% C.L.
Old Area (All)
New Area
Sample Count
1-5
6 - 15
16 - 25
0 20 40
Kilometers > 25
52 MBOMIPA Game Surveys and Participatory Monitoring Programme
Dry Season (October) 1999 Dry Season (November) 2000 Dry Season (October) 2002
Sixth Aerial Game Survey 53
REEDBUCK
Numbers 95% C.L.
Old Area (All)
REWMP 1 - Jul 94 0
REWMP 2 - Nov 94 0
REWMP 3 - Jun 95 0
MBOMIPA 1 - Apr 99 32 107
MBOMIPA 2 - Oct 99 31 91
MBOMIPA3 - May00 0
MBOMIPA4 - Nov 00 10 187
MBOMIPA5 - Apr 01 0
MBOMIPA6 - Oct 02 0
REWMP 3 - Jun 95 0
MBOMIPA 1 - Apr 99 0
MBOMIPA 2 - Oct 99 23 104
MBOMIPA3 - May00 0
MBOMIPA4 - Nov 00 10 190
MBOMIPA5 - Apr 01 0
MBOMIPA6 - Oct 02 0
REWMP 3 - Jun 95 0
MBOMIPA 1 - Apr 99 22 199
MBOMIPA 2 - Oct 99 5 358
MBOMIPA3 - May00 0
MBOMIPA4 - Nov 00 0
MBOMIPA5 - Apr 01 0
MBOMIPA6 - Oct 02 0
New Area
SABLE
Numbers 95% C.L.
Old Area (All)
REWMP 1 - Jul 94
REWMP 2 - Nov 94
REWMP 3 - Jun 95
MBOMIPA 1 - Apr 99 0
MBOMIPA 2 - Oct 99 261 104
MBOMIPA3 - May00 0
MBOMIPA4 - Nov 00 152 103
MBOMIPA5 - Apr 01 379 93
MBOMIPA6 - Oct 02 264 115
Merged estimate 214 88
REWMP 3 - Jun 95
MBOMIPA 1 - Apr 99 0
MBOMIPA 2 - Oct 99 153 159
MBOMIPA3 - May00 0
MBOMIPA4 - Nov 00 50 134
MBOMIPA5 - Apr 01 82 166
MBOMIPA6 - Oct 02 132 188
REWMP 3 - Jun 95
MBOMIPA 1 - Apr 99 0
MBOMIPA 2 - Oct 99 72 203
MBOMIPA3 - May00 0
MBOMIPA4 - Nov 00 101 143
MBOMIPA5 - Apr 01 299 97
MBOMIPA6 - Oct 02 132 137
Merged estimate 116 119
New Area
MBOMIPA 1 - Apr 99 0
MBOMIPA 2 - Oct 99 0
MBOMIPA3 - May00 0
MBOMIPA4 - Nov 00 0
MBOMIPA5 - Apr 01 0
MBOMIPA6 - Oct 02 0
Sixth Aerial Game Survey 55
Sample Count
1-5
6 - 15
16 - 25
0 20 40
Kilometers > 25
56 MBOMIPA Game Surveys and Participatory Monitoring Programme
Dry Season (October) 1999 Dry Season (November) 2000 Dry Season (October) 2002
Sixth Aerial Game Survey 57
SHOATS
Numbers 95% C.L.
Old Area (All)
REWMP 3 - Jun 95 20
MBOMIPA 1 - Apr 99 2,230 191
MBOMIPA 2 - Oct 99 0
MBOMIPA3 - May00 0
MBOMIPA4 - Nov 00 0
MBOMIPA5 - Apr 01 2,569 189
MBOMIPA6 - Oct 02 0
New Area
Sample Count
1-5
6 - 15
16 - 25
0 20 40
Kilometers > 25
Sixth Aerial Game Survey 59
Dry Season (October) 1999 Dry Season (November) 2000 Dry Season (October) 2002
60 MBOMIPA Game Surveys and Participatory Monitoring Programme
WARTHOG
Numbers 95% C.L.
Old Area (All)
New Area
Sample Count
1-5
6 - 15
16 - 25
0 20 40
Kilometers > 25
62 MBOMIPA Game Surveys and Participatory Monitoring Programme
Dry Season (October) 1999 Dry Season (November) 2000 Dry Season (October) 2002
Sixth Aerial Game Survey 63
WATERBUCK
Numbers 95% C.L.
Old Area (All)
REWMP 3 - Jun 95 0
MBOMIPA 1 - Apr 99 51 183
MBOMIPA 2 - Oct 99 47 214
MBOMIPA3 - May00 0
MBOMIPA4 - Nov 00 41 190
MBOMIPA5 - Apr 01 94 163
MBOMIPA6 - Oct 02 41 191
Merged estimate 48 121
New Area
MBOMIPA 1 - Apr 99 0
MBOMIPA 2 - Oct 99 0
MBOMIPA3 - May00 0
MBOMIPA4 - Nov 00 0
MBOMIPA5 - Apr 01 0
MBOMIPA6 - Oct 02 0
64 MBOMIPA Game Surveys and Participatory Monitoring Programme
Sample Count
1-5
6 - 15
16 - 25
0 20 40
Kilometers > 25
Sixth Aerial Game Survey 65
Dry Season (October) 1999 Dry Season (November) 2000 Dry Season (October) 2002
66 MBOMIPA Game Surveys and Participatory Monitoring Programme
ZEBRA
Numbers 95% C.L.
Old Area (All)
New Area
Sample Count
1-5
6 - 15
16 - 25
0 20 40
Kilometers > 25
68 MBOMIPA Game Surveys and Participatory Monitoring Programme
Dry Season (October) 1999 Dry Season (November) 2000 Dry Season (October) 2002
Sixth Aerial Game Survey 69
Between April 1999 and October 2002, a series of six aerial surveys was carried out within the
MBOMIPA study area (Main Report). To all intents and purposes, the operational procedures
adopted for each of the six censuses were strictly comparable. The most important outputs
were population estimates for individual species of wildlife and domestic stock. However, as is
the nature of the method, these estimates demonstrate considerable variation, both within
each census (in terms of confidence limits) and between censuses (marked variation within a
set of population estimates for an individual species) 11. This analysis was therefore undertaken
to gauge the extent to which these two sources of variation could be ascribed to random
(sampling) error rather than to seasonal effects or even to time trends.
Turned around, this F ratio provides a diagnostic tool to test if a series of samples has indeed
been drawn from “the same” population (Table B1). In this Ruaha data set, none of the
calculated F ratios is greater than the critical value of 5.05 (for n-1 = 5 df). This suggests that
sampling error alone is enough to explain the observed variation in population estimates both
within and between surveys.
Table B1: Mean variance (F) ratios for all six surveys
11
Because animals are never distributed evenly within a census zone each sample unit (transect) will vary in the number of animals
that it contains, so that some units will have many animals in them while others will have but few, or perhaps none at all. This means
that the estimate of the population total will depend on the particular distribution of animals that the transects happen to overlay at the
time of survey, for different distributions of the same population will give different results. In other words, for any given population of
animals, the number of alternative estimates that can be obtained from samples of the same size is large. It is this that is known as
“sampling error”, and in spite of the contrary implications of the word “error”, this is not something that can be corrected through better
practice, as it is a fundamental attribute of sampling methodology.
Sixth Aerial Game Survey 70
As a practical consequence, the six estimates of population size for each species can be
merged together to give a single, more precise, estimate (it is best practice here to weight each
individual estimate inversely proportional to its variance).
2. Seasonal effects
Despite the results in (1) above, it is still worthwhile to investigate any seasonal effects on the
population estimates. Certain species may move into or out of the area on a seasonal basis,
and this may contribute towards the between-survey variation even though the individual
estimates are, statistically speaking, drawn from “the same” populations.
A simple one way analysis of variance (ANOVA) is the appropriate test here, in which the six
estimates for each species are tested for a seasonal (“dry” or “wet”) effect (Table B2). Although
in general the dry season estimates are less than the wet season estimates, only with zebra is
the difference statistically significant. Any seasonal effect is accordingly quite weak, and even
pooling all wildlife and all livestock together fails to show anything of significance.
3. Time Trends
It is equally worthwhile to investigate whether a time effect within the survey programme is
contributing towards the between-survey variation in population estimates. Species might be
drifting out of the area due to decreasing rainfall, or through increasing human activity, or
perhaps into the area in response to good rainfall or to avoid human activity.
Sixth Aerial Game Survey 71
OLS (ordinary least squares) regression is the appropriate diagnostic tool here, at least in the
first instance to see if there is anything of interest going on (Table B3). The results are
intriguing. Looking at wildlife first, seven of the eight species demonstrate a consistent
decrease in their population estimates over the 43 months of the survey programme (negative t
values on the time variable), elephant being the sole exception. Although none of the individual
trends is statistically significant, the t values for giraffe, buffalo and zebra are interestingly large.
Something is definitely going on, and when all wildlife are pooled together (using dummy
variables to control for different species effects) the time trend is statistically significant
(p<=0.02). Transforming the population estimates to log10 values intensifies the influence of the
time trend (p<=0.002). Overall, the populations of wildlife in the study area seem to be
decreasing by some 0.7% per month.
However, the time trend is not particularly strong (Table B4). By pooling all the wildlife data into
the first three surveys versus the second three, it is seen that the time variable (TIMEONE)
accounts for at most 6% of the total between-survey variation.
In contrast, the livestock data show no such time effect. Although the t values of all the time
variables are positive, the effect is weak and cannot even be shown by pooling all livestock
together.
4. Summary
In conclusion, the best model from a practical viewpoint is to treat the six estimates for each
species as independent samples from the same population, and pool the results to obtain a
more precise, single estimate (weighting each individual result appropriately). Seasonal effects
are effectively absent (apart from zebra) and may indeed reflect nothing more than changing
visibility bias. However, with wildlife, there does appear to be a weak but consistent time effect,
with populations showing a general decrease over the 43 months of the survey programme.
The decrease is not particularly strong, but must be borne in mind when considering
management options.
Sixth Aerial Game Survey 72
Analysis of Variance