TANZANIAN COASTAL FORESTS: STATUS AND BIOLOGICAL DIVERSITY A. Dickinson', N.D. Burgess? & G.P. Clarke? The Society for Environmental Exploration (Frontier), studio 210 Thames House, 566 Cable Street, London, UK. RSPB (Africa Section), The Lodge, Sandy, Beds., UK. 3. Frontier mnzania Coastal Forest Research Progranme, PO Box 9473, Dar os Sal ABSTRACT The Coastal Forests of Tanzania comprise small geographically isolated forest remnants which may have been isolated from other forest blocks in Africa for the past c. 30 million years and hence have high biological diversity with substantial levels of endemism. Details are presented of 34 known and 9 probable localities supporting Goastal Forest in Tanzania. Individual localities are generally less than 20km?, and there is probably 350-500kn# of forest remaining. Many single localities support endemic or near endemic plants and animals, and there are several genera endemic to the forest type as a whole. The known biological diversity of various forests in terms of plants and birds is presented, along with an attempt to compare biological diversity of sites with the area of forest. The forests provide an extreme example for the conservation of biologically rich habitat "islands".INTRODUCTION Dry forest occurs as an archipelago of small forested "islands" throughout the moister habitats of coastal Tanzania, to within 80km of the coast. The forests are part of a vegetation mosaic along with grassland, woodland and bushland, which has been classified as the Zanzibar-Inhambane Regional Mosaic by White (1983). This phytogeographical area extends along a narrow strip of the East African coast from southern Somalia to Maputo in Mozambique, and contains species and genera that are endemic to that area. White has identified three major lowland forest types within this narrow coastal strip: Zanzibar-Inhambane Lowland Rainforest, which is found at the base of the Eastern Arc montane blocks, and which receives a high annual orographic rainfall; and Zanzibar-Inhambane Scrub Forest and Zanzibar- Inhambane Undifferentiated Forest, both usually present on low hills, riverine areas and occasionally on the coast. The latter two types are drier than the rainforest and undergo a marked seasonal dry period. They can be remarkably variable in terms of species dominance and vegetation structure, the changes occurring over just a few hundred metres. Although altitudinal variation within the forests is low, other factors such as soils, slope, geology and aspect evidently have a significant influence over the forest habitats. The Undifferentiated Forests and Scrub Forests (hereafter referred to as the Coastal Forests) contain large numbers of endemic and near-endemic species and genera, ie. species with distribution limited to one or more of the forests. They have been recognised as a centre for endemic birds by the International Council for Bird Preservation (ICBP, 1992) and up to 10% of the plants are endemic to the forest type. For this level of endemism to develop, the forest must have been both present in this area for a substantial period of time and isolated from other forest blocks for much of this period. HISTORY OF THE FORESTS Kingdon (1990) claims that one reason for this high number of endemic and near-endemic plants occurring along the Indian Ocean Coastal Belt has been the constant climatic influence of the Indian Ocean. This has varied little in temperature during the last 30 million years, remaining a "warm tropical sea" In comparison the Atlantic Ocean cooled by 4-50 during the Ice-Ages which occurred during this period. Evidence for the continuous presence of forest in this area comes from the disjunct distributions of many plant species and genera, which are only recorded from the East African Coastal Forests and the Guineo-Congolian forests of West Africa, but not in the intervening 1000km. Failure of any of these species to disperse across non-forested areas would indicate that a continuous band of forest once stretched between these two forest blocks, and that the Coastal Forests at least pre-date their separation during the upwelling of the Central Tanganyikan plateau some 35 million years ago. There is botanical evidence to suggest that there have been at least two periods of connection between the coastal rain forests (which include the Afromontane forests) and the Zaire forest; a relatively recent one and a more ancient one which would explain the similarities at the generic level (Hamilton, 1973). Kingdon believes that the forests may have last been connected as recently as 3 million years ago. Endemic species and genera that are further limited in distribution to just one or two of the forests indicate long periods of isolation not only from the other major forest blocks, but also between coastal forests.The remaining patches of coastal forest largely occur in moister areas such as hilltops, riverine and groundwater areas. Hills do not have to be high at all this close to the coast to collect significant orographic precipitation. Dryer forest can be found in steep gullies and other sites protected from fire. As the last ice age reached its zenith 18,000 - 20,000 years ago, the world climate is currently enjoying a warm and wet phase, and excluding man's impact on the environment, the world's forests are currently at a maximum of interglacial extent. The occurrence of dry forest in fire-protected sites may therefore indicate that the isolation of the forests is largely due to the anthropogenic fire regime, which started before 50,000 B.C. in Africa, and later due to clearance for agriculture. The current complex distribution of forest patches could represent what were once the wetter areas within much larger and less fragmented areas of dry forest. The Coastal Forests in Tanzania are now restricted to a few dozen sites of jess than 500 km? total extent (Burgess et al., 1992), and no more than 2000km? occurs in the whole of East Africa. Most, possibly all, are currently under grave threat from over-exploitation due to illegal logging, charcoal extraction and clearance for cultivation. Most forests lie within Forest Reserves and owe their current existence to this fact, but insufficient resources are available for policing, and at current rates of conversion there will be no areas of primary vegetation left in fifty years. No coherent management strategy exists for the forests, and unlike Tanzania's other major forest types, no example of Coastal Forest is preserved in a National Park. ASSESSING THE DISTRIBUTION OF THE FORESTS Even in the late 1980s, the area of Tanzania's remaining Coastal Forests, both within and without Forest Reserves was poorly known, and their status and biological value could only be estimated from the scattered results of collections carried out over the last 100 years. In 1989, the Frontier-Tanzania Coastal Forest Research Programme was initiated between the UK based Society for Environmental Exploration and the University of Dar es Salaam in Tanzania, with the objectives of surveying all areas of Coastal Forest, and producing biological inventories of species in each forest studied. This information was to be used to help decide conservation priorities in terms of the biology of each site. Identification of potential forest sites was carried out with the aid of a map of forest cover for Tanzania (Rodgers et al. 1985) produced by interpreting satellite imagery of the 1970s (Figure 1.). Sites indicated by this map, plus other areas mentioned in scientific literature and by local experts, were ground surveyed to assess habitat types present, their extent and condition, their flora and fauna, and human uses. The revised map generated from these site visits is shown in Figure 2. A total of from 34 to 43 coastal forest sites were found in Tanzania with a total area of 400 to 500 km?, and averaging only 12 km2. The majority of these sites, and over 75-85% of the total area, fall within forest reserves. The current extent of coastal forest is considerably less than that indicated by the satellite images, due both to confusion with dense woodland and to recent clearance. It is however considerably more than had been indicated by scientific literature due to insufficient previous surveying. The number of forest sites has been further raised by this survey through the inclusion of numerous very small areas (3km? down to 0.lkm?). These had been ignored by previous field research, and excluded from the satellite survey as insignificant contributors to total forest area. However they were found to still contain important populations of many rare species.Further results.of the survey are summarised in Table 1. Forest area, legal status and number of recorded rare and endemic plants, vertebrates and invertebrates are given. The latter columns include data from past researchers, and reflect as much trends in the distribution of study effort over the various sites as they do real biological trends. They do correctly indicate known endemism "hot-spots", but are obviously neither complete enough or standardised enough to provide in themselves the basis for assigning conservation priorities. Significantly, in every site surveyed thoroughly so far at least one rare species or exceptional distributional record has been found, even in forests under 3km?. What the table does not illustrate is the great diversity of the forests: variations between numbers of distinct community types occur both within and between forests. Furthermore, the "rare" species of one forest were not necessarily the rare species of another: 80% of species occurring in the "rare" columns in Table 1 were found at only one of the listed localities. This is not due to rare species being more likely to be missed in surveys - species are defined as rare here for having a limited distribution, but may be extremely abundant where they do occur. Such relationships can be more clearly examined by closer analysis of those particular taxonomic groups which can give us good enough data-sets, such as birds and plants. Eighteen of the forests have thorough enough bird samples to make their species lists comparable, and there is also an expanding and very comprehensive list of plants for each forest. In combination, these could provide a way to set real conservation priorities between those coastal forests which have been sampled intensively for both groups. BIRDS AS AN INDICATOR OF CONSERVATION PRIORITY Coastal Forests in East Africa support at least six species of bird which are not found anywhere else, of which four species are regarded as threatened by the International Council for Bird Preservation. There are around 100 other species found in Coastal Forests which are forest dependant. Data on the diversity of bird species within each of the forests studied, and a rarity score of each forest bird assemblage based on a scoring system developed by the International Council for Bird Preservation (Collar and Stuart 1988), has been used to show a relationship between the species richness of the forest bird community and the number of rare bird species present. Figure 3 shows the relationship between bird diversity and forest area. A logarithmic regression analysis has been performed on these data and the line of this relationship is shown. This graph indicates that larger forests support a larger number of forest dependant species in a roughly linear fashion, tapering off in larger forests, which fits well with what one would expect from the theory of island biogeography. However, and more interestingly, there would appear to be a tendency for forests below 10km? in area to support a poor assemblage of forest bird species. This might well imply that forests down to that size are important for the conservation of coastal forest bird species, but below that 10km? many forest bird species will become extinct, leading to a very poor bird assemblage. As further forest sites are investigated, this relationship can be tested more formally.Figure 4 presents the rarity scores of the forest bird assemblages from each of these forests against the area of forest. Again this shows that rarity scores increase with remaining size of forest. However, there also seems to be a tendency for forests below 10km? to support no rarities, and therefore to have low conservation importance. Again, it may be that forests below 10km? are too small to be worthy of consideration for conservation in respect of their avifauna. A further indication of size itself being a prime determinant of a forest's importance for bird conservation is seen when comparing these small Tanzanian forests with the Arabuko-Sokoke forest in Kenya. This coastal forest covers some 400km?, and despite being mostly dryer and scrubbier than the Tanzanian forests it contains many more bird rarities than any Tanzanian coastal forest. It has been ranked as the 6th most important forest for birds in Africa by the ICBP criteria used above (Collar and Stuart 1988). PLANTS AS AN INDICATOR GROUP Between one and two hundred plant species are entirely restricted to the coastal forests out of a total constituent flora of in the order of 2000 species. We can gain some idea of coastal forests’ global significance by making a comparison with the Afromontane forests of the Eastern Arc mountains, whose importance as a centre of diversity and endemism is well known. Afromontane forest also comprises in the order of 2000 species, but out of these over 1000 may be endemic to the forest type (data from Iversen 1991). However, coastal forest covers only around 1/sth the area covered by Afromontane forest, making numbers of endemics per square kilometre of forest roughly equal for both habitats. This comparison is not be exact as published figures for diversity and endemism are not strictly comparable, but it is clear that coastal forests are a prime example of a threatened habitat reduced to a very small size. As with birds, some correlation between plant species richness and forest size is expected. Island biogeography theory suggests that larger habitats may support greater species diversities; in addition to this, plants can have marked preferences for different habitats offered within a forest, and larger forests will tend to include a greater variety of habitats than smaller ones. Figure 5 shows relationships between forest size and numbers of rare and endemic plants found within. For this study "endemics" were defined as species so far recorded from a single site only, and "rare" as those known from only five other sites within the phytogeographical province as defined by the Flora of Tropical East Africa (Milne-Redhead et al, 1952 to 1992). Data sources used included all published and unpublished species lists for Tanzanian coastal forests and the Flora of Tropical East Africa. Again there is a distinct trend for numbers of rare or endemic species to decrease with decrease in forest size, down to about 10km?. Below 10km?, in contrast to the pattern noted for birds, it appears size has little further effect and that significant numbers of rarities may persist in the smallest forest fragments. However, this part of the graph may be showing us doomed populations: much of the diversity of these small forest patches may represent individuals living out their span in forests too small to support long-term reproductively viable populations.When figures for rarities and endemics are separated, the relationships are found to differ: larger size may allow greater endemism, but does not guarantee it. Further data will show whether or not this difference is due to sample error, but the result is not unexpected: rates of species evolution and dispersal will vary between sites of the same size due to different local environmental and biogeographical conditions, and number of endemics would be more sensitive to these factors than number of rarities. CONCLUSIONS Several points arise from this work which may be relevant to efforts to conserve biodiversity elsewhere in the tropics, especially the conservation of fragmented forests: a) Existing available information was not sufficient for the drawing up of correct conservation priorities; a substantial scientific field-survey programme was necessary to fill in the gaps in information to allow identification of all sites and correct prioritising of sites. b) Satellite imagery was valuable tool for identifying potential forest sites, but these identifications needed checking on the ground (and no indication of species richness or rarity of a forest can be gained from imagery). ¢) Data so far indicate species differences between forests are such that any hopes of saving all constituent species in a representative sample of forests may be futile; there will always be some species not represented in any sample of forests taken. 4) Even the tiniest forest fragments may contain important populations of rare species, and may even have their own endemic species. Conservation Programmes should not base conservation priories purely on arbitrary criteria such as size, growth form or condition. e) The forests are comprised of a large number of distinctive community types. Even if a number of forests could be protected so as to avoid extinction of a further single coastal forest species, these would not necessarily preserve examples of all the unique types of forest present, nor therefore of the full range of interactions between species. £) A range of indicator groups should be used for the setting of conservation priorities, as factors such as forest size, history and microclimate are of different significance for different groups. Though obtaining data on patterns of rarity and endemism for invertebrates requires great effort, we may be committing a serious oversight in ignoring it: for example the millipede community has been found to differ at the genus level between coastal forests. Making conservation judgements based solely on plant and bird diversity could have serious implications for invertebrate diversity. g) Ideally, further reduction in size of any forest should not be considered until a management strategy has been worked out for al of the forests. Size reduction should not be considered for any forest intended for ultimate conservation, especially below the 10km? limit. h) Active habitat regeneration, including re-afforestation, should be considered as a high priority for sites reduced below the "safe size" of 10km? and so at significant risk of suffering species loss or even extinction.i) Repair of larger forests may be warranted where they have been degraded to the point of being vulnerable to fire or permanent community alteration. 4) Reconstruction of areas of dryer forests as buffer zones around the wetter existing forests should be considered; there is evidence such dry coastal forest used to be very extensive and contained species of its own. k) A large overlap of species, and often exchange of individuals, between the forest and the surrounding woodland or other habitat occurs for all taxa. Assessment of the nature and importance of these linkages for both forest and woodland is necessary for the formulation of reliable long-term management policies. LITERATURE CITED Burgess, N.D., Mwasumbi, L.B., Hawthorne, W.J., Dickinson, A., Dogget, R.A. 1992. Preliminary Assessment of the Distribution, Status and Biological Importance of Coastal Forests in Tanzania. Biological Conservation. 62. 205-218. Burgess, N.D., Dickinson, A. and Payne, N.H.. Tanzanian coastal forests: a poorly known, globally important but highly threatened resource. Oryx (in press). Collar, N.J. and Stuart, S.N. 1988. Key forests for threatened birds in Africa. International Council for Bird Preservation: Monograph No. 3. ICBP, Cambridge. Hamilton, A.C. 1973. The History of Vegetation. in Lind & Morrison East African Vegetation. Iversen, §.T. 1991. The Usambara Mountains, NE Tanzania: Phytogeography of the vascular plant flora. Sym. Bot. Upsaliensis XXIK:3 Milne-Redhead, E., et al (Eds.) 1952 to 1992. Flora of Tropical East Africa. A.A. Balkema, Rotterdam. Rodgers, W.A., Mizray, W. & Shishira, E.K. 1985. The extent of Forest Cover in Tanzania using satellite imagery. University of Dar es Salaam Institute of Resource Assessment, Research Paper No. 12. White, F. 1983. The Vegetation of Africa. UNESCO, Paris.FIGURE 1: Estimated extent of Tanzanian coastal forest from interpretation of satellite images of the 1970s. Dashed lines delineate forest of other types Nzi84 (eg. Afromontane); Solid areas = forests Hatching = thicket. (From Rodgers, W.A., Mizray, W. and Shishra, E.K. 1985. The extent of forest cover in Tanzania using satellite imagery. University of Dar es Salaam Institute of Resource Assessment, Research Paper, No.12.) a Novtran 6°.FIGURE Known and probable. (*), coastal forest localities in ‘Tanzania. TANGA REGION Muheza District 1 Horohoro 2 lulu it 3 Tongwe 4 Rwani/Makinyumbi * Tanga District 5 Tanga Limestone 6 vanbe Is. * Handeni, District 7 Gendagenda 8 Hgambo Pangani District 9 Msubuewe, 10 Pangani River TL Nkwaja DAR ES SALAAM REGION Kinondoni District 12” Pande COAST REGION Bagamoyo District 13 Zaraninge(Kiono} 14 Ruvu North F.R. Kisarawe District. Pugu and Kazimzumbwi Ruvu South Rufiji District 19° Mchungu 20 Namakutwa-Nyamuete 21 Kiwengoma Mafia District 22 Eastern Seaboard 23° Kilindoni LINDI REGION Kilwa District 24 Tong’ ombe 25 Mbinga 26 Mitundembea FR area + 27 Rungo FR area * 28 Nearama NA S FR area 29 Pindiro FR area Lindi District Rondo Litipo Ghitoa Niyangamara Naimba "FR area Ruawa FR area Hatapwa FR area MTWARA REGION Newala District 32 Chilangala 38 Mahuta’ * PEMBA ISLAND Negezi 40 Msitu Mkuu i) Ras Kiuyw ZANZIBAR ISLAND 82 Jozani 44 Ruyund For explanation of FR see footnote 1 to Table 1. L 19° TANZANIA ZANZIBAR 22 23 MAFIA 42 DAR ES SALAAM 494 6°4TABLE 1: Status and summary biological importance of Tanzanian coastal forests Forest Localities Area? Protected? Plants: Inverts: 3 Vertebrates: (aunbers relate to Fig. 1) (kx) Status —_ Indenic Rare Endenic Rare tndenic Rare 1 Horetoro dl oY 7 eee 2 Kilulu HiI1 (Moe) oa) oy - te 3 Tongve 3 00x (12) re 4 vant & Makinyenbi u 8 S25 6 = 5 Yanga Linestone 3 8 yoo oe ew ed 6 Yanbe ts. u 8 er 7 Gendagende 26 bersAS¥R (9519) Tt - - 0 0 8 Mgarbo c. 20 re 9) Msubugve 10-15, r (4) Por. 2 2 e 10 Pangani River 0 soe eee 1 Mkvaja ad - oe ee 12 Pande 1 6 (12) 36 2 + 0 0 13 Kono/taraninge® 0 Ar (219) sou 1 1 os 1 Rave North 1-10 (319) soe ee 1S (Pogu n a (2) 137 sone naty (1) 6 (Rezinzwnbvi 4 1 (49) - 3 1 too S$ 16 Ruva South 38 4 (355) - 1. + 0 2 1 Vikinds 10 r (16) Pot. - 0 o1 1 Kisiju 2 pee ee teen) 19 change 1 1 (10) = 6 2 Loe 20 Nanakutva-lyenuete 0 r (66) - 2 oe ee 21 Kivengoaas 18-25 (a sw 9 moms 2 Mafia Hastern Seaboard = 3 102 om 2 1 aa) 2 Kilindoni Ol a ee) 24 Tongtonbe 10 B (25) soe ee 8 25 Mbinga oS ror (a se 16 Nitandenbea 7.2. area u 2k (85) Se eee 17 Rungo F.R. area u 2 (226) see 28 (Nigatana W FAR. area ‘ 2 (326) - oe oe eee (igarana $ F.R. area 5+ 100k (20) cor ee ; 29 Pindiro FR. ares Se 1008 (118) see 3 Rondo 18-20 100s + (135) ee OT SL Litipo eld los (10) Woe 2 2 ed St Chitoa 1 100 or (8) S25 5 5 2 = 33 Nyanganara ‘ 8 S35 5 5 = 5 Mo Ndinha FR. area 1 2 Ar (13) woe eee 35 Huava FAR, area t ror (29) se 36 Matapua FR. area t 2 2 (165) ce 37 Chilangala el rar (0.1) soe ee 38 Mahuta ‘ 2 Ir (13) sot oe ee 9 geet 15 100k Ir (14) pos yo 40 Msitu uo 13 7 soe ee AL Ras Kinys al o Poe eee 42 Jorani 5 wok ir (3) Lob 10) 3 43 ayant as 8 | Footnotes: 1, Areas vhere given are derived fron ground survey. u = possible presence indicated by satellite photographs but no site visit, reported, The latter uacoafiraed localities are generally assigned nanes relating to the Forest Reserve (FR) they Iie near or vithin, c = eirca (*approxinately*). 2, % = percentage of forest known or thought to fall vithin a protected area. 1 = exact Location of protected area boundaries vith respect to forest uncertain. & = Protective Forest Reserve (no extraction peraitted); r = Productive Forest Reserve (intended for sustainable extraction); ¢ = Gane Reserve (ao extraction peraitted). All Reserves are under the control of Central Goverament (Forestry and Beekeeping Division or Cane Division) unless narked vith "1" indicating Local Goveranent control. Total gevetted area (to the nearest ika2} is given in parentheses. Notable non-forest-dependant species are excluded frou this table. Rare species are those described as such in literature, fadeaics are those known only frou a single, or tvo proxinal forests such 2s Jovani and Muyuni. Fatentheses indicate subspecies. " - " indicates no data available 4, Forest Lies vithin Zaraninge Proposed Forest Reserve. The legel status of this area is undergoing revision. 5. There is local disagreesent over the exact location of District and Forest Reserve boundaries at Kivengoua forest.FIGURE 5. Relationship between numbers of rare and endemic plant species and remaining area of forest. | Number of rare plus endemic plant species against size of coastal forest. (orests with fewer than three hundred specimen identifications are excluded.) 3” 2 40 |. 20 * 9 — oo nm —_—t —_— 10 15 20 Area of Forest (km2) 25‘Ayeanoedses Z2°0 pus p2°0 = 2v ‘OINSGN = Saebs ‘3uVH = Spuowelp : Aa (Zury) IserO,J Jo vary 00's 00°02 o00°St 00°01 00's 000 $$ ~ . er 000 a oO a t Ve a : ° . saa aes = 1 o0"OL | ° a ia | Zz = a ; - 00°02 é t 2. . aa 00°0e E | C n t t _ + —— <—+ 00°0P f Sooo — — — oo'os (‘pepnyoxe ar suoneoyruepr ueuntoeds pospuny een} ue JOMay WIM $}S0J0,]) ‘OZIS JSoIOJ ysuTeSe soroods yue]d ofWopuS puke OIBI JO SIOquINNY 89°0 = dvuFIGURE 3. Relatiénship between number of forest-dependent bird species and size of forest (km2): (A) line regression; (B) logarithmic regression; (C) interpretation. total forest birds against forest area (Q oo 50. 4o total species 30 20 10 aren of forest y = 15.805 + 1.8021x RAZ = 0.767 b&) . 60 30 40 ‘otal species() 30 total species 20 oan 2 total species 40 30 y # 15.805 + 1.8021" RAZ = 0.767 10 == 0 10 20 30 area of forest ¥ = 9.1182 + 29.7774LOG(x) RAZ = 0.640 60 z 40 30Sao) Jo Bale. OPI = 2vd CODE 62 + 28116 =X a0} 40 BAe