Survey Overview The South Pacific Regional Herbarium (SPRH) at the Institute of Applied Science
(IAS), University of the South Pacific coordinated the biodiversity and archaeological
surveys. The surveys were carried out in two phases; 17-26 July, 2012 and 18-26
The focus of the surveys was to map areas of high biodiversity and archaeological
importance, develop monitoring protocols following Climate, Community &
Biodivesity Alliance (CCBA) standards, and later contribute towards a national
biodiversity monitoring protocol for REDD+ projects.
Study Area The REDD+ pilot site is the land encompassed by the mataqali Emalu in the province
of Navosa (Map 1). The Emalu area lies in the climatic transitional zone in central Viti
Levu in the province of Navosa, which is adjacent to five other provinces; Nadroga,
Ba, Namosi, Naitasiri and Serua. The Emalu site covers an area of 7, 347ha,
predominately covered by closed forest.
The Emalu land is under the ownership of the Draubuta landowners, a population of
about 400 people living in more than 30 households in the village of Draubuta. They
constitute five land-owning units, or mataqali, namely:
Mataqali Emalu (Yavusa: Emalu, Tokatoka: Emalu, Duiyabe).
Situated on an alluvial plain, with the Nasa Creek running along the village and
spilling into the upper reaches of the Sigatoka River, accessibility to Draubuta is a
steep 1.5km descent by foot from the nearest dirt road. The main source of income is
through commercial agriculture, mainly kava and taro.
The Emalu land encloses two important catchment areas; (1) the Nasa catchment (on
the drier part of Viti Levu) which drains into the Sigatoka river, and (2) the Mavuvu
and Waikarakarawa catchments (on the wetter part of Viti Levu) discharges into the
The terrain is primarily steep with the highest peak of Mt. Vonolevu having an
elevation of 1,111m. The western side of Emalu forest, which is the drier part of the
transition zone, borders the talasiga grassland. On the eastern side towards Namosi,
which is the wetter part of the transition zone, are the closed rainforests.
The underlying geology of the area is Tholo Plutonic Suite, an intrusive type of rock
which intruded into the Wainimala Group and which is dated to the Upper Miocene
period or later (Rodda, 1967, Rodda, 1976). The alluvial soils from the tributaries in
the Navosa region are classified as Wainibuka sandy clay loam (Twyford and
Wright, 1965, cited in King, 2004). These soils contain high phosphate levels (King,
2004) therefore making it optimum for planting. Rainfall is the major factor
determining the different vegetation systems (requiring different microclimates) and
on average ranges from 2000 to 4000 mm annually (Derrick, 1951)
CHAPTER 2: FLORA, VEGETATION & ECOLOGY Senilolia H. Tuiwawa, Hans Wendt and Marika V. Tuiwawa
2.1 Summary A total of 707 plant taxa were recorded for Emalu, including 286 bryophytes and
lichens, 375 angiosperms, nine gymnosperms, and 35 ferns and fern allies.
Altogether, the vascular and non-vascular taxa recorded from the Emalu site
spanned 182 families and 391 genera. Over a third (39%) of the vascular plant flora
recorded are endemic to Fiji, including 160 species of flowering plants, two fern and
fern allies, and two gymnosperms. Ten taxa were encountered that are important
focal species due to their rarity, botanical significance, very recent discovery in Fiji
and inclusion in the IUCN Red List.
Five principal vegetation types were identified: lowland rainforest, upland rainforest,
cloud forest, dry forest and talasiga/grassland. The lowland rainforest, dry forest and
the associated riparian vegetation were the most heavily impacted, indicated by the
evidence of garden terracing and communal living and can be classified as
anthropogenic primary forest. The upland rainforest and cloud forest were the least
impacted vegetation types and can be described as relatively primary rainforests
with comparatively higher tree species diversity and density.
2.2 Introduction There is literally little or no botanical information available for the Emalu area. This
report is therefore the first documentation of vascular and non-vascular plants of the
Emalu forest. The objectives of this survey were:
To document the range of vegetation types and typical botanical communities
within the study area,
To identify the presence (or potential presence) of species or ecosystems of
national and/or international significance, and,
To assess the susceptibility of the biological communities (in particular plant
communities) to the potential impacts associated with the proposed project.
2.3 Methodology The first phase of the botanical work was carried out within the Tovatova catchment.
The field work assessment began in the open grasslands about 500m upstream from
Navitilevu Village along the Nasa Creek. This was necessary for comparative
purposes towards the incursion of similar forest/habitat types at higher altitudinal
areas within the Emalu boundary.
The second phase of the botanical work was carried out within the Mavuvu
catchment from a basecamp located along Waikarakarawa stream. At
Waikarakarawa, Mavuvu and Mt. Vonolevu, field work began at the lowland
rainforest and progressed towards the cloud forest.
Habitat analysis Prior to the fieldwork an initial assessment of the study area was made using satellite
imagery. It was noted that areas closest to Navitilevu Village were covered in
grassland, and that farming activities were further upstream along the Nasa River.
Bamboo stands were plentiful in areas that have been left as fallow for long periods
of time. Areas beyond these farms were covered with forest. The higher altitude
mountainous landscape towards Mt. Vonolevu (1,111m) was earmarked as an area of
interest. This area would have cloud forest vegetation, normally found at elevations
above 850m, although some have been reported from mountain top from lower
elevations e.g. Mt. Korobaba at 400m elevations (Kirkpatrick and Hassall, 1985).
Cloud forest is a system known to have very unique plant life, and the forest was
expected to be relatively intact.
Flora Survey Opportunistic collections were made of mosses, liverworts, hornworts and lichens on
different substrates. These were packeted for further processing and identification
(Fig. 11, Fig. 12) at the South Pacific Regional Herbarium. Bryophyte collections of
Harris (1967), Fife (2004), Pocs (2004) and Renner (2012) were cross-referenced
including the most recent cryptogram work undertaken for Fiji (von Konrat et al.,
2011, Lumbsch et al., 2011, Soderstrom et al., 2011).
Native plant species, especially those endemic to Fiji, and any species flowering
and/or fruiting, were opportunistically collected and documented whilst trekking
through the area. Additionally plant species observed within the belt transects set up
to quantitatively assess plant density, distribution and diversity within the forest
types, were also documented. Collected specimens were deposited at the South
Pacific Regional Herbarium where verification of specimen identification was later
carried out. In validating the identification, the collection of fertile materials with
special emphasis on the unknown groups i.e. bryophytes and focal species (on the
IUCN Red List) was carried out. The distribution of these taxa within the area
covered was also marked (Map 2) and recommendations for their protection are also
highlighted. Plant names follow those used by Smith (1981, 1979, 1985, 1988, 1991) for
gymnosperms and flowering plants, and those used by Brownlie (1977) and Perrie
and Brownsey (2011) for the pteridophytes (ferns and fern allies).
Vegetation community structure In documenting the range of principal vegetation types and forest or habitat
communities, the topographic and forest cover maps for Fiji, as well as satellite
images were initially used to identify representations of the various plant
communities. A reconnaissance of the area was carried out during which sites were
selected for location of transects and plots.
Quantitative assessment of the communities in different forest types was carried out
using 10m x 10m plots along a 100m transect, a methodology used previously in
other sites in Fiji (Mueller-Dombois and Fosberg, 1998, Tuiwawa, 1999).
Plots were used to:
assess the presence and absence of focal species,
characterise associated vegetation communities with each principal vegetation
confirm boundaries between biological communities encountered.
Within each plot, every tree with a diameter at breast height (dbh) greater or equal to
5cm was measured, identified and recorded. The bole height, crown height and
width were estimated for each tree enumerated. Ground cover vegetation was
described, canopy cover estimated and in addition, the epiphytic flora recorded.
Where feasible GPS locations and photographs of the vegetation were taken.
Habitat characterization Habitat characterisations for forested areas relied on a number of sources of
information, and was undertaken to produce a stratified hierarchical habitat
classification. These sources of information and levels of classification were as
plot data to determine vegetation community structure,
principal vegetation types (Mueller-Dombois and Fosberg, 1998),
1:50,000 topographic map indicating terrain features, and
plot data to determine impact status and likely dominant species and their
The non-forested areas included open country (rivers, open riparian areas, roads,
villages and settlements) and agricultural land (subsistence plantations, commercial
farms, pastures and fallow land). These non-forested areas were not assessed in
detail but were briefly described and highlighted in the vegetation map for the
project (Map 3 and Map 4). The assessment of the vegetation was focused more on
forested area then on non-forested areas.
For the forest or habitat typing process the most prominent topographical feature of
the forested area was used and categorised as follows:
Slope - forested area found on slopes with a gradient ranging from 10 to 85
Ridge top - forested area found on top or along a ridge or mountain range.
The width of such ridges could range from a few centimetres up to 20m, with
an unlimited length.
Flat - forested areas with a gradient ranging from 0 to 10 degrees. These areas
also included raised river flats and flood plains.
In addition to the forest typing, a system of assigning an impact status to each area
was developed (Appendix 4). Whilst rudimentary, the aim of this system was to
provide guidance on which areas of forest have previously been subjected to
disturbance. The categories for the impact were as follows:
Low - Primary forest in which there is little or no evidence of disturbance;
Medium - Secondary forest that is recovering and displays some of the
ecological complexity and function associated with a secondary and primary
forest type or a transition forest type; and
High - Secondary forest that shows signs that the disturbance is recent and on-
2.4 Results and discussion 2.4.1 Diversity of non-vascular plants (bryophytes) and lichens A total of 286 non-vascular plant taxa were collected, recorded and assigned to 72
families, and 133 genera (Appendix 1). The largest families for each of the main
bryophyte groups were Neckeraceae (for mosses), Lejeuneaceae (for liverworts),
Anthocerotaceae (for hornworts) and Graphidaceae (algal symbionts or lichens).
Bryophytes and lichens in the area are typical features in the various systems given
the high elevation. The diversity and density is, at a glance, expectedly low given
that the study area is on the dry side of the island. The collections made will all
contribute to documented range extensions of these species. These records will be
incorporated in the planned publication of the Flora of Bryophytes for the Fiji
Islands, currently in preparation. These non-vascular groups of plants are also
fundamental to climate change research.
2.4.2 Diversity of vascular plants A total of 421 taxa were recorded, of which 395 taxa were identified to species level
and 36 taxa to genus level. Of the total taxa recorded there were 375 angiosperms
(315 dicots and 60 monocots), 35 ferns and fern allies, and nine gymnosperms. For
those taxa identified 350 were native, which comprised 185 indigenous species (145
angiosperms, 32 ferns and eight gymnosperms) and 165 endemic species (160
angiosperms, two ferns and two gymnosperms). A total of 71 exotic species (48
dicots, 22 monocots and one fern), were also recorded. Ten taxa were encountered all
of which were native and considered important focal species. The full species
checklist is provided in Appendix 2.
2.4.3 Focal Species There were a total of ten species encountered which were considered important due
to their rarity, botanical significance, their very recent discovery in Fiji or their IUCN
Red List status. The locations of some of these ten focal species within the study area
are shown on Map 2, and photographs of them in Fig. 1-Fig. 10.
Acmopyle sahniana Buchh. & N.E.Gray (Fig. 1)
The species is endemic to Fiji and found only on the island of Viti Levu. On this
survey a relatively large and viable population was recorded in the upland
vegetation of the Mavuvu Creek at 600-700m. In Fiji, this species is quite restricted in
distribution with an estimated 150 trees recorded to date from the highlands of
Namosi, Naitasiri and Ra Province and is currently classified as Critically
Endangered on the IUCN Red List (Thomas, 2013). A. sahniana is locally known as
kautabua or drautabua and its greatest threats are from mining or logging
operations. Its occurrence in the lower and upper Mavuvu catchment is a promising
confirmation of a new population of A. sahniana and more importantly, a range
extension of its current distribution.
Degeneria vitiensis I.W.Bailey & A.C.Sm. (Fig. 2, Fig. 3)
This species and the two others within the genus are endemic to Fiji. Degeneria is the
sole genus of the family Degeneriaceae, also a Fiji endemic, and one of the oldest
flowering plant families in the world. D. vitiensis is a timber tree found in lowland
and upland areas on the islands of Viti Levu, Vanua Levu and Taveuni, and is
categorised as Vulnerable on the IUCN Red List (World Conservation Monitoring
Centre, 1998). D. vitiensis is locally known as vavaloa or masiratu and its greatest
threats are from the clearance of forest for agriculture and human habitation. Its
occurrence in the upper parts of the lowland and upland area of the study area (590-
650m) suggests a range extension of its current distribution.
Macodes cf. petola (Fig. 4)
This orchid was not documented in Smith’s Flora Vitiensis Nova (1979), and the closest
occurrence of the genus is in Vanuatu. Its occurrence in Fiji is seemingly the result of
long distance dispersal but given evidence of human habitation in adjacent lower
rainforest, it is likely that the taxon was part of an early introduction to Fiji by early
settlers. This species was collected for the second time in Fiji since 2005 and was
recorded in the upland rainforest vegetation on the slopes adjacent to the ridges, at
650-750m. Further research on its population structure and distribution is required to
confirm the mode of introduction into Fiji’s archipelago.
Nervilia cf. punctata (Fig. 5)
This species of orchid is rare in Fiji, known previously from a single collection on Mt.
Korobaba in 1979. Outside of Fiji it is also known to occur in Borneo and Sumatra.
This terrestrial orchid was recorded on slopes of the lowland rainforest vegetation at
Equisetum ramosissimum (Desf.) subsp. debile (Roxb.) Hauke (Fig. 6)
This species of fern ally is not common elsewhere apart from Viti Levu with
extensions eastwards into the tropical Pacific. There are only three records of the
species from Ba, Naitasiri and Serua. It is recorded as common along the edges of
river and creeks. It was recorded in the riparian vegetation off the banks of Nasa
River, at about 500m and is a strong indicator of an intact riparian system.
Metroxylon vitiense (H.Wendl.) Hook. f (Fig. 7, Fig. 8)
This species of palm is endemic to Fiji and is locally referred to as soga. It is common
on the south east of Viti Levu and Vanua Levu but is highly threatened due to its
harvesting for food (palm heart) and for leaves for thatching. Its habitat (swamp) is
targeted for land reclamation both for agriculture development and human
habitation. Very few trees were observed along the river embankments in the lower
Mavuvu River catchment.
Cyphospermatanga (H.E.Moore) H.E.Moore (Fig. 9, Fig. 10)
This palm is endemic to Fiji and is locally referred to as taqwa and is one of the most
threatened palm species in Fiji due to logging and establishment of plantations
(Watling, 2005). Its only other known wild population occurs along the western
slopes of Mt. Tomaniivi (Fiji’s highest mountain, 1323m). The second viable
population, encountered during this survey, lies just outside the eastern Emalu
Alpinia parksii (Gillespie) A.C.Sm.
This species of coarse herb, unlike Alpinia boia, is uncommon. It grows up to 5m high
along river banks and ridges. It becomes more prominent further up along ridges
and slopes. Its leaves are used for thatching and it is locally known as locolocoor boiaboia. Balaka diffusa Hodel
This species of palm is known to be restricted to the Nabukavesi River catchment on
Viti Levu. Its occurrence in the cloud forest of Mt. Vonolevu is a range extension of
the current population. It is an endemic species, locally referred to as balaka, and is
the largest of all balaka in Fiji in terms of its height and girth.
Geanthus cevuga (Seem.) Loesener
This species of ornamental plants is native to Samoa (where it is common) and Fiji
(where it is rare). It was thought previously to only occur in Namosi but recent
findings have confirmed its occurrence in other parts of Fiji as an ornamental. The
leaves are used to scent coconut oil and necklaces and as a food spice. It is locally
referred to as cevuga damu. In Emalu it was found in an old village site growing
alongside sacasaca, molikania and koka trees.
2.4.4 Vegetation community structure Of the nine principal vegetation types recorded for Fiji, five were encountered in the
study area: lowland rainforest, upland rainforest, cloud forest, dry forest and talasiga
grassland. The dry forest refered to here is a mesic forest. All vegetation types except
the talasiga grassland were quantitatively assessed.
The detailed results of the quantitative assessment of plots in these different
vegetation types are given in Appendix 3. In total 136 plots were analysed; 40 in
lowland forest, 41 in upland forest, 44 in dry forest and eleven in cloud forest. Within
each of these vegetation types the plots were distributed over a variety of forest
Vegetation type 1: Lowland rainforest Lowland rainforest is typically found on the windward side of large islands from sea
level to 650m, with annual rainfall of over 2000mm. In Emalu the lowland rainforest
is restricted to the Mavuvu and Waikarakarawa catchments. Overall, the forest in
this principal vegetation type is best described as primary forest. All tree species
recorded from the plots are either endemic or indigenous species and none of the tree
species associated with human habitation was encountered or even observed outside
plots. Stocking of good quality timber tree species is high and so is the size of
merchantable tree species.
Three different forest types were observed and quantified using 40 plots in four
transects. The forest types were characterised as follow:
Ridge top forest type The 20 plots assessed contained an average of 26 (range 9 to 54) trees per plot, and an
average of thirteen (range 7 to 23) species per plot. The most common species was
Garcinia myrtifolia (laubu), which was present in more than 50% of the plots assessed.
The largest individual trees were Decussocarpus vitiensis (amunu) with a dbh of
107cm, followed by Endospermum macrophyllum (kauvula) with a dbh of 95cm and
Calophyllum vitiense (damanu) with a dbh of 82cm. The average dbh was 15cm (range
Overall, the twenty plots assessed had no single dominant species but the
combined biomass (as reflected in the dbh) of D. vitiensis, E. macrophyllum, and C. vitiense gave a relative dominance of 52%.
Slope forest type The seventeen plots assessed had an average of nineteen (range 11 to 28) trees per
plot, and an average of eleven (range 7 to 16) species. Garcinia myrtifolia occurred in
more than 60% of the plots assessed, and was the most common species. The average
dbh of trees in the plots was 15cm (range 5-100cm). The largest tree documented in
the plots was Endospermum macrophyllum with a dbh of 100 cm followed by Storckiella vitiensis (marasa)with a dbh of 78 cm. There was no single dominant species as the
tree sizes were evenly distributed amongst all species, but the combined biomass (as
reflected in the dbh) of E. macrophyllum, S. vitiensis and Calophyllum vitiense gave a
relative dominance of 51%.
River flat forest type The three plots assessed had an average 23 (range 15 to 33) trees per plot, and an
average of fourteen (range 12 to 16) species per plot. The most common tree species
were Garcinia myrtifolia, Garcinia pseudoguttifera and Endiandra gillespiei. The average
dbh of trees in the plots was 18cm (range 5-131cm). The largest tree encountered was
Decussocarpus vitiensis with a dbh of 131cm, and this was also the dominant species
with a relative dominance of 86%.