Article

Integrative taxonomy of the Central African forest chameleon, Kinyongia adolfifriderici (Sauria: Chamaeleonidae), reveals underestimated species diversity in the Albertine Rift

Abstract and Figures

The Albertine Rift (AR) is a centre for vertebrate endemism in Central Africa, yet the mechanisms underlying line-age diversification of the region's fauna remain unresolved. We generated a multilocus molecular phylogeny consisting of two mitochondrial (16S and ND2) and one nuclear (RAG1) gene to reconstruct relationships and examine spatiotemporal diversification patterns in the AR endemic forest chameleon, Kinyongia adolfifriderici (Sternfeld, 1912). This widely distributed species was revealed to be a complex of four genetically distinct and geographically isolated species. Three new species are described based on molecular analyses and morphological examinations. We find that K. rugegensis sp. nov. (Rugege Highlands) and K. tolleyae sp. nov. (Kigezi Highlands) form a well-supported clade, which is sister to K. gyrolepis (Lendu Plateau). Kinyongia itombwensis sp. nov. (Itombwe Plateau) was recovered as sister to K. adolfifriderici (Ituri Rainforest). The phylogeographic patterns we recovered for Kinyongia suggest that speciation stemmed from isolation in forest refugia. Our estimated diversification dates in the Miocene indicate that most species of Kinyongia diverged prior to the aridification of Africa following climate fluctuations during the Pleistocene. Our results highlight the AR as a focal point of diversification for Kinyongia, further elevating the global conservation importance of this region.
Content may be subject to copyright.
© 2017 The Linnean Society of London, Zoological Journal of the Linnean Society, 2017, 181, 400–438 400
Zoological Journal of the Linnean Society, 2017, 181, 400–438. With 10 figures.
Integrative taxonomy of the Central African forest
chameleon, Kinyongia adolfifriderici (Sauria:
Chamaeleonidae), reveals underestimated species
diversity in the Albertine Rift
DANIEL F. HUGHES1*, CHIFUNDERA KUSAMBA2, MATHIAS BEHANGANA3 and
ELI GREENBAUM1
1Department of Biological Sciences, University of Texas at El Paso, El Paso, TX 79968, USA
2Laboratoire d’Herpétologie, Département de Biologie, Centre de Recherche en Sciences Naturelles, Lwiro,
République Démocratique du Congo
3Department of Environmental Sciences, Makerere University, P.O. Box 7298, Kampala, Uganda
Received 4 December 2015; revised 8 December 2016; accepted for publication 15 January 2017
The Albertine Rift (AR) is a centre for vertebrate endemism in Central Africa, yet the mechanisms underlying line-
age diversification of the region’s fauna remain unresolved. We generated a multilocus molecular phylogeny consist-
ing of two mitochondrial (16S and ND2) and one nuclear (RAG1) gene to reconstruct relationships and examine
spatiotemporal diversification patterns in the AR endemic forest chameleon, Kinyongia adolfifriderici (Sternfeld,
1912). This widely distributed species was revealed to be a complex of four genetically distinct and geographically iso-
lated species. Three new species are described based on molecular analyses and morphological examinations. We find
that K. rugegensis sp. nov. (Rugege Highlands) and K. tolleyae sp. nov. (Kigezi Highlands) form a well-supported
clade, which is sister to K. gyrolepis (Lendu Plateau). Kinyongia itombwensis sp. nov. (Itombwe Plateau) was
recovered as sister to K. adolfifriderici (Ituri Rainforest). The phylogeographic patterns we recovered for Kinyongia
suggest that speciation stemmed from isolation in forest refugia. Our estimated diversification dates in the Miocene
indicate that most species of Kinyongia diverged prior to the aridification of Africa following climate fluctuations dur-
ing the Pleistocene. Our results highlight the AR as a focal point of diversification for Kinyongia, further elevating
the global conservation importance of this region.
ADDITIONAL KEYWORDS: biodiversity – biogeography – Burundi – conservation – Democratic Republic of the
Congo – diversification – molecular systematics – new species – phylogeography – Uganda.
INTRODUCTION
The Albertine Rift (AR) represents the western branch
of the East African Rift valley system (Chorowicz,
2005). The modern topography of the East African Rift
started to form c. 30 Mya from volcanic swells in North
and East Africa (Paul, Roberts &White, 2014). The AR
portion of the East African Rift was likely initiated
during the Oligocene (c. 25 Mya) (Roberts et al., 2012).
However, most of the geomorphological changes in the
AR took place in the mid- to late Miocene (15–5 Mya)
(Macgregor, 2015). Volcanism also principally occurred
during the Miocene in the AR (Nonnotte et al., 2008),
and extensive lava flows would have contributed to
landscape modifications (Griffiths, 1993). The AR is not
only geologically unique, it also harbours more endemic
vertebrate species than any other area of similar size
on continental Africa (Plumptre et al., 2007), including
the highest mammalian tropical forest species rich-
ness per unit area on Earth (Demos et al., 2015). The
AR was not identified by Myers et al. (2000) as one of
the original 25 Biodiversity Hotspots. Reassessments
with an African emphasis, however, elevated the AR
into the Eastern Afromontane Hotspot (Brooks et al.,
*Corresponding author. E-mail: dfhughes@min-
ers.utep.edu [Version of Record, published online
20 May 2017; http://zoobank.org/ urn:lsid:zoobank.
org:pub:FAAE0F53-0BED-4D04-BB71-15A392FC9195]
Downloaded from https://academic.oup.com/zoolinnean/article-abstract/181/2/400/3836902
by University of Texas at El Paso user
on 28 November 2017
SYSTEMATICS OF ALBERTINE RIFT FOREST CHAMELEONS (CHAMAELEONIDAE: KINYONGIA) 401
© 2017 The Linnean Society of London, Zoological Journal of the Linnean Society, 2017, 181, 400–438
2004) and to a Global Biodiversity Hotspot (Küper
et al., 2004). Nevertheless, the proposed mechanisms
of speciation or environmental processes that have
sculpted the immense diversity of the region are not
conclusive. Aridification and refugia formation in
response to Pleistocene glaciations have been impli-
cated as drivers of isolation and subsequent lineage
formation among some AR montane taxa, including
small mammals (Demos et al., 2014, 2015), land snails
(Wronski & Hausdorf, 2008; Boxnick et al., 2015), goril-
las (Anthony et al., 2007) and birds (Bowie et al., 2006;
Voelker, Outlaw & Bowie, 2010). In contrast, several
other AR taxa, including frogs (Portillo et al., 2015;
Larson et al., 2016), chameleons (Tolley et al., 2011)
and snakes (Menegon et al., 2014; Greenbaum et al.,
2015) were suggested to have evolved from pre-Pleisto-
cene events such as the reduction of forests and spread
of grasslands across Africa in response to global cool-
ing in the Miocene. Because the AR has a complex oro-
genic history and is one of the most important sites for
biodiversity in Africa (Plumptre et al., 2007), it pro-
vides an ideal opportunity to understand the relative
influences of historic biogeographic events and climate
change on spatiotemporal aspects of speciation.
The forest chameleon genus Kinyongia Tilbury,
Tolley & Branch, 2006 currently contains 20 described
taxa that are distributed in forests across East and
Central Africa (Uetz & Hošek, 2016). Monophyly for
Kinyongia was established using nuclear and mito-
chondrial markers, but unique morphological synapo-
morphies have not been identified (Tilbury et al., 2006).
Currently recognized Kinyongia species were histori-
cally classified under the genus Chamaeleo Laurenti,
1768 and then reallocated to the South African genus
Bradypodion Fitzinger, 1843 by Klaver & Böhme (1986).
This transfer was not based on similarity, but rather
the lack thereof, and the taxonomic rearrangement was
not accepted by Branch (1998). Klaver & Böhme (1986)
acknowledged the lack of morphological synapomor-
phies among species in this group [previously referred
to as the ‘fischeri complex’ (Hillenius, 1959)], and
this heterogeneity – among other reasons (see Tolley
& Herrel, 2013) – influenced the tangled and contro-
versial taxonomic histories of Kinyongia (see Tilbury
et al., 2006) and Bradypodion (see Tolley et al., 2004).
Morphological dissimilarity among Kinyongia species
is exemplified by differences in cranial ornamentation
of this oviparous group, which includes paired rostro-
nasal horns [e.g. K. fischeri (Reichenow, 1887)], a single
blade-like rostral horn [e.g. K. xenorhina (Boulenger,
1901)] or no cranial ornamentation (e.g. K. mulyai
Tilbury & Tolley, 2015). Kinyongia was historically
thought to be sister to the dwarf chameleons (genus
Bradypodion) (e.g. Tolley et al., 2011). However, Tolley
et al. (2013) found forest chameleons (Kinyongia) to
be most closely related to horned chameleons (genus
Trioceros Swainson, 1839) and proposed an ancient
split between these sister genera in the mid-Eocene (c.
45 Mya). Kinyongia is composed of mostly ancient lin-
eages originating in the Oligocene and early Miocene
(Tolley et al., 2011; Tolley & Herrel, 2013). The specia-
tion patterns in Kinyongia recovered by Tolley et al.
(2011) did not reflect recent phylogeographic struc-
turing in response to climate changes during the
Pleistocene, as recovered in some African vertebrate
taxa (e.g. Arctander, Johansen & Coutellec-Vreto,
1999). Rather, stronger genetic signatures in Kinyongia
seemed to stem from the forest dynamics over this time
period, including reductions (Plana, 2004) and diver-
sifications (Couvreur et al., 2008). Tolley et al. (2011)
proposed an allopatric model of speciation through
isolation in forest refugia and supported this model
with high genetic divergence between most sister spe-
cies with a notable absence of sister species occupying
the same mountain block. However, a presumed recent
divergence of sister species of Kinyongia co-occurring
on the Rwenzori Mountains [i.e. K. carpenteri (Parker,
1929) and K. xenorhina] of the AR are in stark contrast
to this model, and more investigation is warranted to
understand the mechanisms underlying speciation
patterns at this geologically young massif [formed c.
2–3 Mya (Kaufmann, Hinderer & Romanov, 2015)].
In general, all species of Kinyongia are restricted to
relict montane or sub-montane forest biomes (Tolley
& Herrel, 2013) and occupy a relatively high elevation
range (1000–3000 m) (Tilbury, 2010). Three genetically
divergent and geographically isolated clades have been
recognized within Kinyongia; one from the AR/Kenya
Highlands and two from the Eastern Arc Mountains
(EAM) of East Africa (Tolley et al., 2011; Tolley,
Townsend & Vences, 2013). Seven species currently
comprise the AR/Kenya Highlands clade, including five
from the AR [K. adolfifriderici (Sternfeld, 1912), K. car-
penteri, K. gyrolepis Greenbaum et al., 2012, K. mulyai
and K. xenorhina] and two from the Kenya Highlands
[K. excubitor (Barbour, 1911) and K. asheorum Nečas
et al., 2009]. Four of the AR species are endemic to the
montane localities of their original descriptions and the
fifth species, K. adolfifriderici, is currently considered
to be widespread throughout the rift (Tilbury, 2010).
Kinyongia adolfifriderici represents the most westerly
species of the genus, extending into the sub-montane
forests of eastern Democratic Republic of the Congo
(DRC) (Tilbury, 2010; Greenbaum et al., 2012a; Tilbury
& Tolley, 2015). The distribution of this species nearly
covers the latitudinal extent of the AR, ranging on either
side of the rift from the Itombwe Plateau in eastern DRC
and forest remnants in Burundi, to the northeastern
extent of the Ituri rainforest in DRC (Tilbury, 2010). As
a result of this widespread distribution and overlapping
range with some protected areas, the IUCN Red List
currently has listed this species as Least Concern (Tolley
Downloaded from https://academic.oup.com/zoolinnean/article-abstract/181/2/400/3836902
by University of Texas at El Paso user
on 28 November 2017
402 D. F. HUGHES ET AL.
© 2017 The Linnean Society of London, Zoological Journal of the Linnean Society, 2017, 181, 400–438
et al., 2014a). Kinyongia adolfifriderici was described by
Sternfeld (1912) from a single adult female specimen
and the type locality was imprecisely given as ‘Irumu-
Mavambi Urwald [jungle]’. According to Greenbaum
et al. (2012a), Irumu and Mavambi (= Mawambi) are
villages in the lowland Ituri rainforest of present-day
northeastern DRC that were sites visited during the
German Central Africa Expedition (1907–1908) led by
Adolphus Frederick, Duke of Mecklenburg (Frederick,
1910). Since its original discovery, few specimens of this
species have been collected because of its high canopy
habits, cryptic coloration and shy behaviour (Tilbury,
2010). Moreover, persistent civil strife within countries
of the AR (e.g. van Reybrouck, 2014) has further discour-
aged biological exploration, and in turn, contributed to
the rarity of this species in museum collections and lack
of basic biological information (Tilbury, 2010).
At least two studies have proposed that K. adolfi-
friderici represents a complex of species (Greenbaum
et al., 2012a; Tilbury & Tolley, 2015). Various searches in
isolated forest patches of East and Central Africa have
revealed distinct species of Kinyongia (e.g. Lutzmann &
Nečas, 2002; Menegon et al., 2009, 2015; Nečas, 2009;
Nečas et al., 2009), including two new species from the
AR (Greenbaum et al., 2012a; Tilbury & Tolley, 2015).
Therefore, we anticipate that additional undescribed
Kinyongia lineages occur in other poorly explored forest
fragments of the AR. To that end, we examined spatial
and temporal aspects of the phylogenetic relationships
among isolated populations of K. adolfifriderici across
the AR to test three hypotheses. We use multilocus gene-
tree and coalescent-based species-tree estimations to
determine whether K. adolfifriderici represents a single
widespread species in the AR or is it rather a complex
of genetically distinct species. We utilize multivariate
statistical analyses for morphometric data on over 20
specimens, including the holotype to determine if the
lineages recovered from the phylogenetic analyses are
supported by morphological differences. We implement
Bayesian dating methods to estimate divergence times
within Kinyongia to determine if climate shifts induced
by Pleistocene glaciation have had an impact on the
timing of speciation within forest chameleons of the AR.
Finally, we describe three new species on the basis of
morphological characters, mtDNA pairwise sequence
divergences, qualitative observations and congruence
across multiple phylogenetic approaches that support
stable hypotheses of independently evolving species.
MATERIAL AND METHODS
Taxon sampling
Fourteen samples of K. adolfifriderici were collected
during field surveys from various forests across
the highlands of the AR from 2008 to 2015. For our
morphological examinations, we incorporated an addi-
tional eight specimens (K. cf. adolfifriderici) from
various museum collections (Appendix 1). Museum
abbreviations follow Sabaj (2016). We also included
the holotype of K. adolfifriderici (ZMB 22709) in
these examinations. For our phylogenetic analyses,
we included 19 out of the 20 currently recognized
Kinyongia species that have published sequences
available in GenBank (Table 1). We excluded the spe-
cies K. asheorum from phylogenetic analyses because
only a single sequence for the mitochondrial fragment
ND2 is currently available in GenBank. Kinyongia
asheorum, from the Nyiro Range in northern Kenya, is
considered to be a member of the AR/Kenya Highlands
clade; however, this phylogenetic placement was based
on analyses of the single mtDNA sequence (Nečas
et al., 2009; Tolley et al., 2013).
Dna exTracTion, amplificaTion
anD sequencing
Tissues were harvested from the liver or hind limb
muscle of chameleons before formalin fixation, and
preserved in 2 mL vials containing 100% ethanol.
Genomic DNA was isolated from these tissue sam-
ples with the Qiagen DNeasy tissue kit (Qiagen Inc.,
Valencia, CA, USA). PCR amplification and cycle
sequencing of two mitochondrial gene fragments were
carried out following standard procedures with the fol-
lowing primers for ND2: L4437b (Macey et al., 1997a)
and H5934 (Macey et al., 1997b), and 16S: L2510 and
H3080 (Palumbi, 1996). A fragment of the nuclear gene
RAG1 was sequenced using primers F118 and R1067
(Matthee, Tilbury & Townsend, 2004). We used 25 µL
PCR reactions with an initial denaturation step of
95 °C for 2 min, followed by denaturation at 95 °C for
35 s, annealing at 50 °C for 35 s and extension at 72 °C
for 95 s, with 4 s added to the extension per cycle for
32 (mitochondrial genes) or 34 (nuclear gene) cycles.
Amplification products were visualized on a 1.5% aga-
rose gel stained with Invitrogen SYBR Safe DNA gel
stain (Thermo Fisher Scientific, Waltham, MA, USA).
Sequencing reactions were purified with Agencourt
CleanSEQ magnetic bead solution (Beckman Coulter
Inc., Brea, CA, USA) and sequenced with an ABI
3130xl automated sequencer at the University of
Texas at El Paso (UTEP) Border Biomedical Research
Center (BBRC) Genomic Analysis Core Facility.
sequence alignmenT anD
phylogeneTic analyses
Twenty-four new sequences were generated from eight
individuals for two mitochondrial markers (16S, ND2)
and one nuclear marker (RAG1). Sequences of two
individuals from Uganda (CAS 201593–94) and an
Downloaded from https://academic.oup.com/zoolinnean/article-abstract/181/2/400/3836902
by University of Texas at El Paso user
on 28 November 2017
SYSTEMATICS OF ALBERTINE RIFT FOREST CHAMELEONS (CHAMAELEONIDAE: KINYONGIA) 403
© 2017 The Linnean Society of London, Zoological Journal of the Linnean Society, 2017, 181, 400–438
Table 1. Species identifications, specimen catalogue numbers, GenBank accession numbers and collecting localities for Kinyongia (ingroup) and Trioceros (out-
group) samples analysed in this study
Species Catalogue no. 16S ND2 RAG1 Locality
Kinyongia rugegensis sp. nov. (T) UTEP 21481 KY292356 KY292364 KY292372 Burundi: Bubanza Province, Mpishi village,
Kibira National Park
Kinyongia rugegensis sp. nov. (H) UTEP 21485 KY292357 KY292365 KY292373 Burundi: Bubanza Province, Mpishi village,
Kibira National Park
Kinyongia rugegensis sp. nov. UTEP 21484 KY292358 KY292366 KY292374 Burundi: Kayanza Province, Rwegura vil-
lage, Kibira National Park
Kinyongia tolleyae sp. nov. (T) UTEP 21486 KY292352 KY292360 KY292368 Uganda: Kabale District, Ruhija village,
Bwindi Impenetrable National Park
Kinyongia tolleyae sp. nov. (T) UTEP 21487 KY292353 KY292361 KY292369 Uganda: Kabale District, Ruhija village,
Bwindi Impenetrable National Park
Kinyongia tolleyae sp. nov. UTEP 21489 KY292354 KY292362 KY292370 Uganda: Kasese District, Ruboni
Community Hotel, Rwenzori Mountains
National Park
Kinyongia tolleyae sp. nov. (T) CAS 201593 DQ923820 EF014304 DQ996659 Uganda: Kabale District, Ruhija village,
Bwindi Impenetrable National Park
Kinyongia tolleyae sp. nov. (T) CAS 201594 GQ221944 GQ221965 N/A Uganda: Kabale District, Ruhija village,
Bwindi Impenetrable National Park
Kinyongia itombwensis sp. nov. (H) UTEP 20371 JN602061 JN602051 JN602056 DRC: South Kivu Province, Bichaka vil-
lage, Itombwe Plateau
Kinyongia itombwensis sp. nov. UTEP 21480 KY292351 KY292359 KY292367 DRC: South Kivu Province, Miki village,
Itombwe Plateau
Kinyongia adolfifriderici (T) UTEP 21491 KY292355 KY292363 KY292371 DRC: Oriental Province, Loki village,
Ituri rainforest
Kinyongia boehmei (T) BM 29 GQ221942 GQ221963 GQ221953 Kenya: Taita Hills
Kinyongia boehmei (T) JM 2946 GQ221948 GQ221969 GQ221958 Kenya: Taita Hills
Kinyongia carpenteri (T) CT 346 DQ923822 EF014306 FR716622 Uganda: Rwenzori
National Park
Kinyongia carpenteri UTEP 20370 JN602058 JN602048 JN602053 DRC: North Kivu Province, western slope
of Ruwenzori Mountains, Mount Teye
Kinyongia excubitor (T) CT 209 DQ923823 EF014307 DQ996661 Kenya: Mount Kenya
Kinyongia fischeri (T) CT 334 DQ923829 EF014313 DQ996662 Tanzania: Nguru Mountains
Kinyongia fischeri (T) MTSN 8490 GQ221951 GQ221971 GQ221960 Tanzania: Nguru Mountains
Kinyongia gyrolepis (T) UTEP 20339 JN602062 JN602052 JN602057 DRC: Orientale Province, Aboro village,
Lendu Plateau
Kinyongia gyrolepis (T) UTEP 20342 JN602055 JN602050 JN602060 DRC: Orientale Province, Aboro village,
Lendu Plateau
Kinyongia magomberae MTSN 8218 GQ221950 GQ221970 GQ221959 Tanzania: Udzungwa Mountains
Downloaded from https://academic.oup.com/zoolinnean/article-abstract/181/2/400/3836902
by University of Texas at El Paso user
on 28 November 2017
404 D. F. HUGHES ET AL.
© 2017 The Linnean Society of London, Zoological Journal of the Linnean Society, 2017, 181, 400–438
Species Catalogue no. 16S ND2 RAG1 Locality
Kinyongia magomberae (H) MTSN 8492 GQ221952 GQ221972 GQ221961 Tanzania: Magombera Forest
Kinyongia matschiei (T) CAS 168852 FR716605 FR716641 FR716626 Tanzania: East Usambara Mountains
Kinyongia matschiei (T) CT 105 GQ221946 GQ221967 GQ221956 Tanzania: East Usambara Mountains
Kinyongia multituberculata (T) CT 111 GQ221947 GQ221968 GQ221957 Tanzania: West Usambara Mountains
Kinyongia mulyai (H) CT 426 KM589402 KM589404 KM589403 DRC: Katanga Province, Mount Nzawa
Kinyongia msuyae (H) MTSN9374 LN997635 LN997645 N/A Tanzania: Livingstone Mountains
Kinyongia msuyae (T) MTSN9375 LN997636 LN997646 N/A Tanzania: Livingstone Mountains
Kinyongia oxyrhina (T) CT 192 DQ923831 EF014315 DQ996669 Tanzania: Uluguru Mountains
Kinyongia oxyrhina (T) CT 193 DQ923832 EF014316 DQ996670 Tanzania: Uluguru Mountains
Kinyongia tavetana (T) CT 113 DQ991233 FJ717801 DQ996671 Tanzania: Mount Kilimanjaro
Kinyongia tavetana CT 207 DQ923833 EF014317 DQ996672 Tanzania: Mount Meru
Kinyongia tenuis (T) CAS 168917 DQ923834 EF014318 HQ130628 Tanzania: East Usambara Mountains
Kinyongia tenuis (T) CT 103 DQ923835 EF014319 DQ996673 Tanzania: East Usambara Mountains
Kinyongia uluguruensis (T) CT 189 DQ923825 EF014309 DQ996667 Tanzania: Uluguru Mountains
Kinyongia uluguruensis (T) CT 191 DQ923826 EF014310 DQ996666 Tanzania: Uluguru Mountains
Kinyongia uthmoelleri CT 151 DQ923836 EF014320 DQ996674 Tanzania: South Pare Mountains
Kinyongia uthmoelleri (T) CT 339 DQ923837 EF014321 DQ996675 Tanzania: Mount Hanang
Kinyongia vanheygeni SCHP-08-R-50 LN997640 LN997650 N/A Tanzania: Poroto Mountains
Kinyongia vanheygeni SCHP-08-R-91 LN997641 LN997651 N/A Tanzania: Poroto Mountains
Kinyongia vosseleri (T) CAS 168921 GQ221943 GQ221964 GQ221954 Tanzania: East Usambara Mountains
Kinyongia vosseleri (T) CT 104 GQ221945 GQ221966 GQ221955 Tanzania: East Usambara Mountains
Kinyongia xenorhina (T) CT 350 DQ923838 EF014322 DQ996676 Uganda: Rwenzori Mountains National
Park
Kinyongia xenorhina (T) CT 351 DQ923839 EF014323 DQ996677 Uganda: Rwenzori Mountains National
Park
Trioceros feae (T) CAS 207681 FJ717767 AF448749 AF448749 Equatorial Guinea: Bioko Island
Trioceros goetzei CT 050 FJ717768 FJ717791 FJ746603 Malawi: Nyika Plateau
Trioceros johnstoni CAS 201596 DQ923812 EF014298 DQ996650 Uganda: Kabale District
Note: Newly generated sequences are indicated with bold type. Institutional abbreviations follow Sabaj (2016). DRC = Democratic Republic of the Congo; T = topotype; H = holotype.
Table 1. Continued
individual from the Itombwe Plateau (DRC) (UTEP
20371) were published previously (Greenbaum et al.,
2012a). New sequences were deposited in GenBank
(Table 1). Outgroup samples included three species
of horned chameleons (Trioceros) (Tolley et al., 2013)
(Table 1). Chromatograph data were interpreted using
SEQMAN (Swindell & Plasterer, 1997). Alignments for
each gene were generated using MUSCLE 3.6 (Edgar,
2004) in MESQUITE 3.04 (Maddison & Maddison,
2015). Manual adjustments and editing were carried
out in MACCLADE 4.08 (Maddison & Maddison, 2005).
A hypervariable region in the 16S gene fragment con-
sisting of 44 base pairs was cut out prior to phyloge-
netic analyses because of an ambiguous alignment.
Phylogenetic analyses were initially conducted on sin-
gle-gene data sets. The resulting individual gene-trees
revealed nearly identical topologies and thus the fol-
lowing analyses were conducted on the concatenated
data set. Maximum likelihood (ML) analyses of con-
catenated data were conducted with the GTRGAMMA
model in RAXML 7.2.6 (Stamatakis, 2006). All param-
eters were estimated and a random starting tree
was used. Support values for clades inferred by ML
analyses were assessed with the rapid bootstrap algo-
rithm with 1000 replicates (Stamatakis, Hoover &
Rougemont, 2008). Bayesian inference (BI) analyses
were conducted in MRBAYES 3.1 (Huelsenbeck &
Ronquist, 2001; Ronquist & Huelsenbeck, 2003). Our
model included seven data partitions, including a sin-
gle partition for 16S and three independent partitions
for each codon position for the protein-coding genes
ND2 and RAG1. Concatenated data sets were parti-
tioned identically for ML and BI analyses. The Akaike
Information Criterion (AIC) in PARTITIONFINDER
1.1.0 (Lanfear et al., 2012) was used to establish the
best model of evolution for the nuclear and each of
the mitochondrial fragments. The selected models of
evolution were used for all BI analyses, but in cases
where the model selected in PARTITIONFINDER
was not available in MRBAYES, the least restrictive
model (GTR) was implemented. Bayesian analyses
were conducted with random starting trees, run for
20 million generations, and Markov chains were sam-
pled every 1000 generations. To verify that multiple
runs converged, AWTY (Nylander et al., 2008) was
used. Burn-in was set at 25%, and thus 5000 of the
initial trees were discarded. Phylogenies were visual-
ized using FIGTREE 1.3.1 (Rambaut & Drummond,
2009). Bayesian posterior probabilities 95% (Hillis &
Bull, 1993; Alfaro, Zoller & Lutzoni, 2003) and boot-
strap values 70% (Felsenstein, 1981, 1985) were con-
sidered as strong support. Net sequence divergences
(uncorrected p-distances) between Kinyongia lineages
for each marker were estimated using MEGA 6.0.5
(Tamura et al., 2013).
Downloaded from https://academic.oup.com/zoolinnean/article-abstract/181/2/400/3836902
by University of Texas at El Paso user
on 28 November 2017
SYSTEMATICS OF ALBERTINE RIFT FOREST CHAMELEONS (CHAMAELEONIDAE: KINYONGIA) 405
© 2017 The Linnean Society of London, Zoological Journal of the Linnean Society, 2017, 181, 400–438
individual from the Itombwe Plateau (DRC) (UTEP
20371) were published previously (Greenbaum et al.,
2012a). New sequences were deposited in GenBank
(Table 1). Outgroup samples included three species
of horned chameleons (Trioceros) (Tolley et al., 2013)
(Table 1). Chromatograph data were interpreted using
SEQMAN (Swindell & Plasterer, 1997). Alignments for
each gene were generated using MUSCLE 3.6 (Edgar,
2004) in MESQUITE 3.04 (Maddison & Maddison,
2015). Manual adjustments and editing were carried
out in MACCLADE 4.08 (Maddison & Maddison, 2005).
A hypervariable region in the 16S gene fragment con-
sisting of 44 base pairs was cut out prior to phyloge-
netic analyses because of an ambiguous alignment.
Phylogenetic analyses were initially conducted on sin-
gle-gene data sets. The resulting individual gene-trees
revealed nearly identical topologies and thus the fol-
lowing analyses were conducted on the concatenated
data set. Maximum likelihood (ML) analyses of con-
catenated data were conducted with the GTRGAMMA
model in RAXML 7.2.6 (Stamatakis, 2006). All param-
eters were estimated and a random starting tree
was used. Support values for clades inferred by ML
analyses were assessed with the rapid bootstrap algo-
rithm with 1000 replicates (Stamatakis, Hoover &
Rougemont, 2008). Bayesian inference (BI) analyses
were conducted in MRBAYES 3.1 (Huelsenbeck &
Ronquist, 2001; Ronquist & Huelsenbeck, 2003). Our
model included seven data partitions, including a sin-
gle partition for 16S and three independent partitions
for each codon position for the protein-coding genes
ND2 and RAG1. Concatenated data sets were parti-
tioned identically for ML and BI analyses. The Akaike
Information Criterion (AIC) in PARTITIONFINDER
1.1.0 (Lanfear et al., 2012) was used to establish the
best model of evolution for the nuclear and each of
the mitochondrial fragments. The selected models of
evolution were used for all BI analyses, but in cases
where the model selected in PARTITIONFINDER
was not available in MRBAYES, the least restrictive
model (GTR) was implemented. Bayesian analyses
were conducted with random starting trees, run for
20 million generations, and Markov chains were sam-
pled every 1000 generations. To verify that multiple
runs converged, AWTY (Nylander et al., 2008) was
used. Burn-in was set at 25%, and thus 5000 of the
initial trees were discarded. Phylogenies were visual-
ized using FIGTREE 1.3.1 (Rambaut & Drummond,
2009). Bayesian posterior probabilities 95% (Hillis &
Bull, 1993; Alfaro, Zoller & Lutzoni, 2003) and boot-
strap values 70% (Felsenstein, 1981, 1985) were con-
sidered as strong support. Net sequence divergences
(uncorrected p-distances) between Kinyongia lineages
for each marker were estimated using MEGA 6.0.5
(Tamura et al., 2013).
species-Tree esTimaTion
In accordance with integrative taxonomy, whereby tax-
onomists should present different lines of evidence to
support a stable hypothesis that a population is evolv-
ing independently (Padial et al., 2010), and because the
underlying species tree can be different than gene trees
(Maddison, 1997), we implemented a multi-coalescent
model to estimate a species tree for the AR Kinyongia
clade in *BEAST 1.8 (Drummond et al., 2012). The
program assumes lineage sorting is the main source of
inconsistency between gene trees and the species tree,
it does not require an outgroup (Heled & Drummond,
2010), and necessitates the prior assignment of individ-
uals to presumed species (Bell et al., 2015). Our species
assignments were based on a combination of morpho-
logical characters and well-supported, geographically
isolated lineages recovered in the concatenated ML
and BI gene trees (Figs 1, 2). Double peaks in the
nuclear marker (RAG1) were rare and most were eas-
ily remedied at the chromatogram stage in SEQMAN
because one allele exhibited a much stronger signal
than the other (unequal heights) (Fontaneto, Flot &
Tang, 2015). Nevertheless, to account for the possibil-
ity of heterozygous individuals in the nuclear data set,
haplotypes for RAG1 were phased using PHASE 2.1.1
(Stephens & Donnelly, 2003) in DNASP 5.1 (Librado
& Rozas, 2009). Haplotypes with probabilities lower
than 0.7 were excluded from the species-delimitation
analysis (Harrigan, Mazza & Sorenson, 2008). Phased
sequences were used for the subsequent analysis
in *BEAST. Models of sequence evolution were cho-
sen using the AIC in PARTITIONFINDER. We only
included samples with complete sequence data for all
loci. We specified unlinked site, clock and tree mod-
els, and implemented a Yule process tree prior as this
analysis largely investigates interspecific relation-
ships. The analysis was run for 50 million generations,
sampling every 1000 generations. Multiple independ-
ent analyses were run to confirm results produced the
same topology. We discarded the initial 25% of trees
as burn-in. Convergence was determined from histo-
grams, trace plots and Effective Sample Size (ESS)
values with TRACER 1.5 (Rambaut & Drummond,
2007). Bayesian posterior probabilities 95% were con-
sidered as strong support.
Estimates of species trees can be validated with
tree-based coalescent approaches (e.g. Niemiller,
Near & Fitzpatrick, 2012). By far the most popu-
lar of the Bayesian tree-based methods is Bayesian
Phylogenetics and Phylogeography (BPP) (Yang &
Rannala, 2010). However, BPP requires a fixed, user-
specified guide tree, which was recognized by Leac
et al. (2014) as the most obvious pitfall to BPP because
inaccuracies in the guide tree can result in artifi-
cial increases in genetic divergence between sister
Downloaded from https://academic.oup.com/zoolinnean/article-abstract/181/2/400/3836902
by University of Texas at El Paso user
on 28 November 2017
406 D. F. HUGHES ET AL.
© 2017 The Linnean Society of London, Zoological Journal of the Linnean Society, 2017, 181, 400–438
lineages, and consequently, over delimitation of species
(i.e. false-positives) (Carstens et al., 2013). For exam-
ple, inappropriate guide trees used by Leaché & Fujita
(2010) resulted in biased support for incorrect mod-
els, and in turn, inflated taxonomic conclusions (see
Bauer et al., 2010). Recently, Olave, Solà & Knowles
(2014) identified several significant problems associ-
ated with the sensitivity of tree-based species delimi-
tation analyses to mistakes with upstream analyses
(i.e. guide trees for BPP). We align with Carstens et al.
(2013) who suggested that results from these analyses
should be interpreted with caution and thus we have
refrained from implementing a tree-based validation
approach until these concerns are fully resolved.
Divergence DaTing
We used the Bayesian program BEAST 1.8 (Drummond
et al., 2012) to estimate divergence dates within
Kinyongia. We implemented an uncorrelated log-
normal relaxed clock model with an estimated clock
rate to allow for rate heterogeneity among lineages
(Drummond et al., 2006). We used a Yule process tree
prior (pure birth) on our multilocus data set because
this prior is best suited for phylogenies describing the
relationships between different species and assumes a
constant speciation rate. Multiple independent anal-
yses were run to confirm results produced the same
topology. Analyses were run for 50 million generations,
sampling every 1000 generations. TRACER was used
to confirm stationarity and adequate ESS of the pos-
terior probabilities (>200 for each estimated param-
eter). We discarded the first 25% of trees as burn-in.
Parameter values from the posterior probabilities
on the maximum clade credibility tree were summa-
rized using the program TREEANNOTATOR 1.7.5
(Drummond et al., 2012). Bayesian posterior probabili-
ties 95% were considered as strong support.
For the dated tree analysis, we included recent rep-
resentatives of more distantly related groups as well as
representative species from all chameleon genera (at
least three species per genus when available) (Table
S1). In some instances, hybrid sequences composed
of different species from the same genus or a closely
related genus (based on Pyron, Burbrink & Wiens,
2013) were used for outgroup samples and these alter-
nate species are indicated in Table S1. A total of 14
nodes were constrained for divergence dating and
most dates for calibration purposes were adopted from
Tolley, Townsend, and Vences (2013). Primary (fossil)
calibrations were placed on a total of nine nodes in the
tree corresponding to some of the oldest known fossils
of lepidosaurian taxa (Table 2). Secondary calibra-
tions were placed on a total of five nodes on the tree to
achieve temporal congruence with the most complete
time-calibrated chameleon phylogeny published to
date – including over 90% of all named species (Tolley
et al., 2013) (Table 2). For each calibration, we used a
translated log-normal distribution, with an offset equal
to the age of the fossil or estimated internal node split.
The results of an initial dated analysis depicted a tree
topology that did not closely reflect the interspecific
relationships among chameleon genera recovered by
Tolley et al. (2013). Therefore, topological constraints
(i.e. enforced monophyly) were imposed on three clades
(Calumma + Furcifer; Bradypodion + Nadzikambia;
Trioceros + Kinyongia) based on the intergeneric rela-
tionships recovered in the family-level phylogeny by
Tolley et al. (2013).
morphological analyses
Specimens examined for this study were preserved
in 10% buffered formalin in the field and transferred
to 70% ethanol for long-term storage in the UTEP
Biodiversity Collections. Other specimens that were
examined morphologically are presented in Appendix
1. Morphometric data were recorded from preserved
specimens with vernier callipers to the nearest 0.1 mm
with the aid of a stereomicroscope. Colour descrip-
tions are based on colour photographs in life, personal
observations and field notes. Sex was determined by
internal examination of gonads, everted hemipenes or
the presence of hemipenal bulges distal to the vent.
Drawings of everted hemipenes were conducted with
the aid of an illuminated stereomicroscope. Hemipenal
terminology follows Klaver & Böhme (1986).
Measurements (±0.1 mm) were taken from the right
side of the body. Morphometric and meristic data, and
their associated abbreviations were modified from
Branch & Tolley (2010) and Greenbaum et al. (2012a):
snout–vent length (SVL) from tip of snout to anterior
edge of vent; tail length (TL) from tip of tail to poste-
rior edge of vent; total length (ToL) from tip of snout
to tip of tail; head length (HL) from superior tip of
casque to tip of snout; head width (HW) measured at
widest point just posterior to eyes; head height (HH)
from rictus (i.e. commissure) of jaw to superior tip of
casque; mouth length (ML) from tip of rostral to rictus;
casque–eye length (CE) measured diagonally from pos-
terior margin of orbit to superior tip of casque; snout
length (SL) from tip of snout to anterior margin of
orbit; eye diameter (ED) measured horizontally at cen-
tre of eye; cranial crest gap (CC) measured across the
crown between raised supraorbital crests at mid-eye;
inter-limb length (IL) from axillary to inguinal attach-
ments of limbs; forelimb length (FLL) from elbow to
wrist; hind limb length (HLL) from knee to heel and
three meristic characters, including conical tubercles
of dorsal crest (CTD), upper labials (UL) to posterior
Downloaded from https://academic.oup.com/zoolinnean/article-abstract/181/2/400/3836902
by University of Texas at El Paso user
on 28 November 2017
SYSTEMATICS OF ALBERTINE RIFT FOREST CHAMELEONS (CHAMAELEONIDAE: KINYONGIA) 407
© 2017 The Linnean Society of London, Zoological Journal of the Linnean Society, 2017, 181, 400–438
margin of orbit and lower labials (LL) to posterior
margin of orbit. Statistical comparisons of selected
measurements and counts were conducted with a
two-tailed t-test. To avoid potential problems with the
use of ratios in statistics, all measurement data were
analysed in an analysis of covariance (ANCOVA), with
body size (SVL) as a covariate (Packard & Boardman,
1999). These analyses were conducted in MINITAB 16
(Minitab Statistical Software, State College, PA, USA).
To examine whether Kinyongia from these popula-
tions exhibit morphological differences, ten continu-
ous morphological measurements were used (HL,
HW, HH, ML, CE, SL, ED, CC, FLL and HLL). Two
variables (TL and IL) skewed the results of an initial
principal components analysis (PCA) and were thus
omitted from the following analyses. These two vari-
ables possess potentially meaningful implications
regarding differences between sexes for TL (males
tend to have longer tails) and gravid females with
extended abdomens inflated IL measurements (com-
pared to males of similar SVL). These ten mensu-
ral characters were size corrected using SVL as a
covariate and the residuals were included in a PCA.
All variables had communalities (>0.5). A varimax
rotation was used, PCs with eigenvalues >1.0 were
extracted, and the resulting PC scores were saved.
These PC scores were then used as input variables
for a multivariate analysis of variance (MANOVA)
with the species as the fixed factor. Post hoc pair-
wise comparisons were made using Tukey’s Honest
Significant Difference test and the Bonferroni test.
These multivariate analyses were carried out in
SPSS 22 (IBM SPSS Statistics for Windows, Armonk,
NY, USA).
Table 2. Divergence-date priors for primary (top) and secondary (bottom) calibrations. Node numbers correspond to those
indicated in Figure 4
PRIMARY CALIBRATIONS
Node TL zero-offset
(mean, SD)
Median (95% CI) Source
1 238 (1.4, 0.7) 242 (239.4–249.6) Fossil rhynchocephalian from mid-Triassic
(Jones et al., 2013)
2 161 (1.8, 1.0) 167 (162.2–192.3) Stem scincomorph Balnealacerta from mid-
Jurassic (Evans, 1998)
3 110 (1.8, 1.3) 116 (110.7–161.3) Stem teiids (e.g. Ptilodon) from early Cretaceous
(Winkler, Murry & Jacobs, 1990; Nydam & Cifelli, 2002)
4 61 (1.6, 0.8) 66.1 (62.4–80) Fossil amphisbaenian Plesiorhineura tsentasi
from mid-Paleocene (Sullivan, 1985)
5 128 (1.0, 0.5) 130.7 (129.2–134.2) Fossil lizard Dalinghosaurus longidigitus from
early Cretaceous (Evans & Wang, 2005)
6 70 (1.8, 1.0) 76.1 (71.2–101.3) Fossil anguid Odaxosaurus from late Cretaceous
(Sullivan & Lucas, 1996)
7 70 (1.8, 1.0) 76.1 (71.2–101.3) Stem acrodont iguanian clade Priscagaminae
from late Cretaceous (Keqin & Norell, 2000)
8 70 (1.2, 1.9) 73.3 (70.2–145.6) Fossil pleurodont iguanian Saichangurvel from
late Cretaceous (Conrad & Norell, 2007)
9 99 (1.0, 0.5) 101.7 (100.2–105.2) Stem chameleon from Albian–Cenomanian
boundary, Cretaceous (Daza et al., 2016)
SECONDARY CALIBRATIONS
Node TL zero-offset
(mean, SD)
Median (95% CI) Source
10 62.8 (1.0, 0.5) 64.8 (63.3–68.3) Node 1 by codon in Table S4 of Tolley et al. (2013)
11 51.2 (1.0, 0.5) 53.9 (52.4–57.4) Node 2 by codon in Table S4 of Tolley et al. (2013)
12 47.5 (1.0, 0.5) 50.2 (48.7–53.7) Node 3 by codon in Table S4 of Tolley et al. (2013)
13 45.6 (1.0, 0.5) 48.3 (46.8–51.8) Node 4 by codon in Table S4 of Tolley et al. (2013)
14 33.4 (1.0, 0.5) 36.1 (34.6–39.6) Node X by codon in Table S4 of Tolley et al. (2013)
The translated log-normal (TL) zero-offset is presented in millions of years ago (Mya), parameter values (mean and standard deviation) follow in
parentheses and posterior (calculated) ages are presented as median with 95% confidence interval in parentheses.
Downloaded from https://academic.oup.com/zoolinnean/article-abstract/181/2/400/3836902
by University of Texas at El Paso user
on 28 November 2017
408 D. F. HUGHES ET AL.
© 2017 The Linnean Society of London, Zoological Journal of the Linnean Society, 2017, 181, 400–438
RESULTS
molecular phylogeneTics, species-Tree
inference anD sequence Divergence
A total of 2118 bp was obtained from three loci for each
of eight individuals of K. cf. adolfifriderici (16S: 441 bp;
ND2: 856 bp; RAG1: 821 bp). There were no gaps in the
alignments of any of the three loci after we omitted a
hypervariable region consisting of 44 bp from the 16S
ribosomal gene. Using AIC in PARTITIONFINDER,
we determined that the most appropriate substi-
tution models were GTR+G for 16S; GTR+I+G for
ND2 first codon position, HKY+G for ND2 second
codon position, TIM+G for ND2 third codon position;
HKY+G for RAG1 first and second codon positions,
HKY+I for RAG1 third codon position. Relationships
among Kinyongia species were reconstructed utiliz-
ing the concatenated data set and the same topology
was recovered using BI and ML methods (Fig. 1). The
three geographically distinct clades from Tolley et al.
(2011) were recovered as monophyletic (i.e. AR/Kenya
Highlands, EAM North and EAM South) (Fig. 1). The
recently described species K. mulyai endemic to Mount
Nzawa in the southern AR was recovered with strong
support as sister to the Rwenzori Massif endemics,
K. xenorhina and K. carpenteri (Fig. 1). Four distinct
lineages representing geographically isolated popu-
lations of K. cf. adolfifriderici were recovered with
strong support (Fig. 2). The populations from south-
western Uganda (Rwenzori Mountains + Bwindi
Impenetrable National Parks) formed a distinct clade
that was most closely related to a clade from northern
Burundi (Kibira National Park), whereas a clade from
the Itombwe Plateau (eastern DRC) was recovered in
a sister relationship with a sample from the Ituri rain-
forest (northeastern DRC) (Figs 1, 2). The placement
of K. gyrolepis from the Lendu Plateau in northeastern
DRC within the K. adolfifriderici clade renders the lat-
ter species paraphyletic.
The species-tree analysis recovered identical clade
topology to our concatenated gene-tree analyses. The
AR Kinyongia clade was strongly supported (Fig. 3). The
K. adolfifriderici clade was recovered with strong support
and in an identical topology to the gene-tree analyses.
Kinyongia gyrolepis was nested within the K. adolfifri-
derici clade, supporting the paraphyletic relationship
recovered in our multilocus gene tree. The four distinct
lineages of K. cf. adolfifriderici were strongly supported
and the sister relationships among them upheld from
the gene-tree estimations (Fig. 3). Only the placement
of K. mulyai was equivocal in this analysis (Fig. 3). The
results of this coalescent-based species-tree inference
further support the recognition of three undescribed,
novel lineages of K. cf. adolfifriderici from the AR.
Pairwise sequence divergences (uncorrected p-dis-
tances) between the undescribed lineages were
generally high and comparable to currently recog-
nized Kinyongia species endemic to the AR (Table S2).
For the ND2 locus, p-distances between the western
Uganda clade and the Burundi and Itombwe clades
ranged from 5.9 to 6.8% and from 7.1 to 9.8%, respec-
tively (Fig. 1). P-distances between the Burundi clade
and the Itombwe clade for ND2 ranged from 9.8 to
10.5% (Fig. 1). Lastly, p-distance ranges for this locus
between undescribed clades and the topotypic Ituri lin-
eage were also high: 6.1–6.5% for Itombwe; 8.2–10.6%
for Uganda and 10.3–10.4% for Burundi (Table S2).
DaTing esTimaTes
Results from the calibrated dating analysis indicate
that Kinyongia diverged from Trioceros in the Eocene,
the three major Kinyongia clades emerged around
the Eocene–Oligocene boundary and major lineage
diversification within Kinyongia took place during
the Miocene (Fig. 4). Of the 22 distinct Kinyongia
lineages recovered in this Bayesian analysis, 18
diverged during the Miocene, three diverged during
the Oligocene and one diverged during the Eocene.
Most AR Kinyongia lineages likely diverged during
the Miocene, as evidenced by estimated mean diver-
gence dates within this epoch (Fig. 4). An estimated
divergence date at the Oligocene–Miocene boundary
around 23.39 Mya [17.77–29.6 Mya, 95% highest pos-
terior densities (HPD)] was given for the split between
the Kenya Highlands (K. excubitor) clade and the AR
clade (Fig. 4). A basal divergence of AR Kinyongia
was estimated in the early Miocene at approximately
17.93 Mya (13.58–23.03 Mya, HPD) (Fig. 4). The spe-
cies K. mulyai from the southern AR diverged from
the Rwenzori Massif endemics (K. carpenteri and K.
xenorhina) around 14.35 Mya (9.63–19.35 Mya, HPD).
The two Rwenzori Massif endemic species were esti-
mated to have diverged from each other in the late
Miocene around 7.17 Mya (4.1–10.83 Mya, HPD). The
root divergence of all lineages in the K. adolfifrider-
ici clade was given in the mid-Miocene at 11.45 Mya
(8.38–15.3 Mya, HPD). Divergence between K. gyro-
lepis and the Burundi + Uganda clade was dated in the
late Miocene around 7.93 Mya (5.28–11.12 Mya, HPD).
Topotypic K. adolfifriderici diverged from the Itombwe
clade also in the late Miocene at 6.37 Mya (3.57–9.78
Mya, HPD). A split between the Burundi and Uganda
clades was estimated to occur at the Miocene–Pliocene
boundary around 5.05 Mya (2.89–7.63 Mya, HPD).
morphological analyses
Analysis of ten continuous characters extracted three
principal components which accounted for 73.9% of the
variation in the data set (Table S3). The first component
(PC1) loaded highest for cranial crest gap (CC) and head
Downloaded from https://academic.oup.com/zoolinnean/article-abstract/181/2/400/3836902
by University of Texas at El Paso user
on 28 November 2017
SYSTEMATICS OF ALBERTINE RIFT FOREST CHAMELEONS (CHAMAELEONIDAE: KINYONGIA) 409
© 2017 The Linnean Society of London, Zoological Journal of the Linnean Society, 2017, 181, 400–438
Figure 1. Maximum-likelihood phylogeny of the African forest chameleon genus Kinyongia. Nodes with strong support from both maximum-likelihood and
Bayesian inference analyses are indicated by filled circles (bootstrap 70% + posterior probability 0.95), and nodes with support from only maximum-likelihood
analyses by open circles. Colour-coded rectangles correspond to the Albertine Rift endemic species as follows (in descending order from top of phylogeny): yellow – K.
cf. adolfifriderici (Kibira – Burundi) = K. rugegensis sp. nov.; purple – K. cf. adolfifriderici (Bwindi + Rwenzori – Uganda) = K. tolleyae sp. nov.; blue – K. gyro-
lepis; red – K. adolfifriderici; dark green – K. cf. adolfifriderici (Itombwe – DRC) = K. itombwensis sp. nov.; pink – K. xenorhina; light green – K. carpenteri; orange
K. mulyai. This colour scheme is retained throughout all figures where applicable. Uncorrected p-distances for the ND2 marker are given as a range for the three
new species (top). The map depicting the northern (dark grey) and southern (light grey) montane regions associated with the Eastern Arc Mountains was modified
from Platts et al. (2011).
Downloaded from https://academic.oup.com/zoolinnean/article-abstract/181/2/400/3836902
by University of Texas at El Paso user
on 28 November 2017
410 D. F. HUGHES ET AL.
© 2017 The Linnean Society of London, Zoological Journal of the Linnean Society, 2017, 181, 400–438
length (HL), the second (PC2) for snout length (SL) and
mouth length (ML), and the third (PC3) for head height
(HH) and eye diameter (ED). The MANOVA showed
a significant difference between all species for PC1
(F3,18 = 4.07, P = 0.023). Post hoc pairwise comparisons
showed differences between topotypic K. adolfifriderici
and the Burundi specimens for PC1 (P < 0.05), and a
near significant value between Burundi and Itombwe
specimens for PC1 (P = 0.07). The scoreplot of the first
two PCs regressed against body size (SVL) indicated
that there is significant overlap for these continuous
measurements, but some population clustering was
discernable (i.e. Burundi and Itombwe) (Fig. 5). Overall,
the morphological characters were conservative across
Figure 2. Elevation map of the Albertine Rift (Central Africa) showing sampling localities of forest chameleons
(Kinyongia) used in this study. Photographs of representative individuals for each species are displayed on right; image
for K. mulyai of the holotype (PEM-R 19199) adapted from Tilbury & Tolley (2015) with permission from Colin R. Tilbury;
image for K. adolfifriderici is a lateral view of the preserved holotype (ZMB 22709) taken by Frank Tillack; photographs
of K. xenorhina and K. cf. adolfifriderici (Bwindi + Rwenzori – Uganda) = K. tolleyae sp. nov. were taken by DFH; and
photographs of K. gyrolepis, K. carpenteri, K. cf. adolfifriderici (Itombwe – DRC) = K. itombwensis sp. nov. and K. cf.
adolfifriderici (Kibira – Burundi) = K. rugegensis sp. nov. were taken by EG.
Downloaded from https://academic.oup.com/zoolinnean/article-abstract/181/2/400/3836902
by University of Texas at El Paso user
on 28 November 2017
SYSTEMATICS OF ALBERTINE RIFT FOREST CHAMELEONS (CHAMAELEONIDAE: KINYONGIA) 411
© 2017 The Linnean Society of London, Zoological Journal of the Linnean Society, 2017, 181, 400–438
the isolated populations, yet there seem to be some dif-
ferences between populations in head dimensions and
other measurements detailed in the following.
Burundi samples had a significantly larger aver-
age SVL (t = 2.44, d.f. = 8, P = 0.04), and more upper
(t = 4.48, d.f. = 8, P = 0.002) and lower (t = 3.11, d.f. = 8,
P = 0.014) labial counts than topotypic Ituri samples.
Burundi samples had significantly more upper labial
counts (t = 2.6, d.f. = 12, P = 0.023) than Uganda sam-
ples. Burundi samples had significantly more conical
tubercles on the dorsal crest than Itombwe samples
(t = 4.56, d.f. = 6, P = 0.004). Uganda samples had sig-
nificantly more upper (t = 2.56, d.f. = 12, P = 0.025) and
lower (t = 2.59, d.f. = 12, P = 0.023) labial counts than
topotypic Ituri samples. Itombwe samples had signifi-
cantly more lower labial counts than topotypic Ituri
samples (t = 2.86, d.f. = 6, P = 0.029).
Size-corrected ANCOVA analyses detected a signifi-
cant difference between Uganda and topotypic Ituri
populations for SL, with Uganda samples larger than
those from Ituri (F1,13 = 14.11, P = 0.003) (Table 3) .
Moreover, a significant difference was found between
Uganda and Itombwe samples for SL (F1,11 = 6.09, P =
0.036), and values at significance for CE (F1,11 = 5.08,
P = 0.05), FLL (F1,11 = 5.05, P = 0.05) and approach-
ing significance for HLL (F1,11 = 4.36, P = 0.06), with
Uganda samples having larger mean measurements
than Itombwe samples for these four variables (Table
3). No size-corrected differences were detected between
Burundi samples and topotypic Ituri samples. Size-
corrected ANCOVA analyses detected significant dif-
ferences between Burundi and Itombwe samples for
FLL (F1,7 = 55.45, P = 0.001) and HLL (F1,7 = 13.83, P
= 0.014), with Burundi samples having larger limbs
than Itombwe samples (Table 3). Also, a value at sig-
nificance between Burundi samples and Uganda sam-
ples was found for head width (HW) (F1,13 = 4.65, P =
0.05), with Burundi samples larger than Uganda ones
(Table 3). No size-corrected differences were detected
between Itombwe samples and topotypic samples from
Ituri. Given the small sample sizes, these results are
preliminary and more analyses based on larger sam-
ples are warranted to determine if these differences
reflect reliable characters. Regardless, these results
demonstrate that there are multiple significant mor-
phological differences between the populations.
Taxonomy
The presumably topotypic K. adolfifriderici genetic
sample from Loki village is located in the Ituri
rainforest, Orientale Province, DRC. This site is
c. 75 km (straight line distance) from present-day
Irumu village, which is one of the sites provided by
Sternfeld (1912) as the type locality for this spe-
cies (see Introduction section). Also, the sample was
obtained at 1408 m elevation, which is well within
the appropriate elevation range for the species
(Tilbury, 2010). With putative topotypic genetic mate-
rial for K. adolfifriderici, we are confident that the
three geographically separated, genetically divergent
lineages in the AR represent new species (Itombwe
Plateau in southeastern DRC; Kibira National Park
in northern Burundi; Bwindi Impenetrable and
Rwenzori Mountains National Parks in western
Uganda) (Fig. 2). Based on the combined molecular
Figure 3. *BEAST species-tree inference for combined data set of Kinyongia species within the clade endemic to the Albertine
Rift (in descending order from top of phylogeny): K. mulyai (n = 1), K. carpenteri (n = 2), K. xenorhina (n = 2), K. cf. adolfifri-
derici (Itombwe – DRC) = K. itombwensis sp. nov. (n = 2), K. adolfifriderici (n = 1), K. gyrolepis (n = 2), K. cf. adolfifriderici
(Kibira – Burundi) = K. rugegensis sp. nov. (n = 3) and K. cf. adolfifriderici (Bwindi + Rwenzori – Uganda) = K. tolleyae sp.
nov. (n = 4). Nodes with Bayesian inference posterior probability values 95% are denoted by filled circles.
Downloaded from https://academic.oup.com/zoolinnean/article-abstract/181/2/400/3836902
by University of Texas at El Paso user
on 28 November 2017
412 D. F. HUGHES ET AL.
© 2017 The Linnean Society of London, Zoological Journal of the Linnean Society, 2017, 181, 400–438
and morphological results presented above, and
qualitative differences explained in the following, we
describe three new species that were previously con-
sidered to be populations of K. adolfifriderici.
DESCRIPTIONS OF THREE NEW SPECIES
family chamaeleoniDae gray, 1825
genus Kinyongia Tilbury, Tolley & branch,
2006
Kinyongia rugegensis sp. nov.
(figs 1, 2, 3, 4, 6, 7a, 10; Tables 3, 4)
urn:lsid:zoobank.org:act:4906DDBE-FB3C-4424-
8ABB-ADF0A76468F0
Figure 4. Bayesian chronogram of Chamaeleonidae. The genus Kinyongia (clade of interest) is expanded for greater detail
(left). Posterior probabilities 95% are denoted by filled circles adjacent to nodes. Numbers above nodes denote mean high-
est posterior densities (HPD) and blue bars at nodes represent 95% HPD. Fossil-calibrated nodes are indicated with encir-
cled red numbers and secondary calibrated nodes with encircled light blue numbers.
Figure 5. Scatter plot of the first two principal components
extracted of morphological characters for species from the
Kinyongia adolfifriderici group. The holotype for K. adolfi-
friderici (ZMB 22709) is indicated with a white star.
Downloaded from https://academic.oup.com/zoolinnean/article-abstract/181/2/400/3836902
by University of Texas at El Paso user
on 28 November 2017
SYSTEMATICS OF ALBERTINE RIFT FOREST CHAMELEONS (CHAMAELEONIDAE: KINYONGIA) 413
© 2017 The Linnean Society of London, Zoological Journal of the Linnean Society, 2017, 181, 400–438
Chamaeleo adolfi-fridericiFischer & Hinkel, 1992:
fig. 110 – Photograph in life and record for Nyungwe
forest, Rwanda.
Bradypodion adolfi-fridericiHinkel, 1993: Record for
Cyamudongo forest, Rwanda.
Common name: Rugege Highlands forest chameleon.
Holotype: UTEP 21485 (field no. ELI 1156), adult
female, BURUNDI, Bubanza Province, near Kibira
National Park, Mpishi village, 03°411.064S
29°294.02E, 1660 m elevation, 20 December 2011, col-
lected by E. Greenbaum, C. Kusamba, M.M. Aristote,
and W.M. Muninga (Fig. 6A).
Paratopotypes: Same collection details as holotype, one
adult male, UTEP 21481 (field no. ELI 1155) (Fig. 6B),
and another adult male, UTEP 21482 (field no. ELI
1238), collected on 23 December 2011 (Fig. 6C).
Paratypes: One adult female, UTEP 21483 (field no. ELI
1220), BURUNDI, Bubanza Province, Kibira National
Park, Mpishi village, 03°342.372S 29°2936.348E,
1986 m elevation, 22 December 2011, collected by
same collectors of holotype; one adult female, UTEP
21484 (field no. ELI 1256), BURUNDI, Kayanza
Province, Kibira National Park, near Rwegura village,
02°5620.292S 29°2954.78E, 2130 m elevation, 25
December 2011, collected by E. Greenbaum, M.M.
Aristote, and W.M. Muninga (Fig. 6D).
Diagnosis: Kinyongia rugegensis sp. nov. can be distin-
guished from all other Kinyongia species by the following
combination of traits: (1) lack of rostro-nasal ornamen-
tation in both sexes; (2) moderate body size (mean
SVL = 55.9 mm); (3) anterior dorsal keel with 8–10 coni-
cal tubercles; (4) a slightly elevated casque that tapers
posteriorly to a prominent apex; (5) absence of a gular
and ventral crest; (6) 16–18 upper and 15–17 lower labi-
als; (7) generally flat shape of the upper casque; (8) tail
length longer than SVL in both sexes; (9) indistinct pari-
etal crest with slightly raised tubercles; (10) background
body coloration in adult females generally green to yel-
low-green with darker pigmented regions on the flanks
and tail; background body coloration in adult males
generally brown with tan and yellow speckling on the
flanks; (11) interstitial skin between the tubercles of the
body generally black for both sexes; (12) a light brown
stripe passes through the middle of the eye and extends
from the canthal ridge to the temporal crest; (13) top of
the head is typically a darker green/brown colour than
elsewhere; (14) the gular region is distinctly lighter in
colour, with a combination of green, white and tan.
Table 3. Summary of meristic and mensural characters in adult type specimens of Kinyongia rugegensis sp. nov.,
K. tolleyae sp. nov., K. itombwensis sp. nov. and examined material of topotypic K. adolfifriderici (including the
holotype)
Kinyongia rugegensis
sp. nov. Kibira Forest NP,
Burundi n = 5
(2M, 3F)
Kinyongia tolleyae sp. nov.
Bwindi + Rwenzori NPs, Uganda
n = 9
(2M, 7F)
Kinyongia itombwensis
sp. nov. Itombwe Plateau,
DRC n = 3
(1M, 2F)
Kinyongia adolfifriderici Ituri
rainforest, DRC
n = 5
(1M, 4F)*
SVL 55.9 ± 2.2 (52.8–58.7) 56.6 ± 5.7 (48.5–66.2) 51.1 ± 4.4 (46.1–54.8) 52.1 ± 2.7 (47.9–54.9)
TL 74.1 ± 10.8 (65.6–90.2) 69.4 ± 5.7 (64.0–75.6) 65.4 ± 2.2 (63.8–67.9) 66.4 ± 8.5 (56.4–76.3)
ToL 130.1 ± 12.2 (118.4–148.9) 125.9 ± 9.9 (112.7–141.4) 116.5 ± 5.1 (110.6–120.2) 117.1 ± 11.3 (104.3–131.2)
HL 16.6 ± 1.1 (15.4–18.2) 16.1 ± 0.9 (14.8–17.3) 15.1 ± 0.8 (14.4–15.9) 15.2 ± 0.7 (14.1–15.9)
HW 8.3 ± 0.5 (7.8–8.9) 7.9 ± 0.4 (7.3–8.6) 7.7 ± 0.7 (7.0–8.5) 7.6 ± 0.4 (7.2–8.2)
HH 9.8 ± 0.7 (8.9–10.5) 10.2 ± 0.8 (9.0–11.8) 9.1 ± 0.5 (8.7–9.6) 9.2 ± 0.8 (8.2–10.0)
ML 11.6 ± 1.0 (10.2–12.9) 12.3 ± 1.2 (10.3–13.6) 10.9 ± 1.2 (9.8–12.2) 11.8 ± 1.1 (10.1–13.1)
CE 7.1 ± 0.6 (6.5–7.9) 6.7 ± 0.3 (6.2–7.2) 6.1 ± 0.3 (5.8–6.4) 6.2 ± 0.4 (5.6–6.6)
SL 5.5 ± 0.5 (4.9–6.4) 5.7 ± 0.3 (5.1–6.0) 5.0 ± 0.4 (4.7–5.4) 4.9 ± 0.3 (4.4–5.3)
ED 5.5 ± 0.6 (4.9–6.3) 5.2 ± 0.2 (4.9–5.6) 4.9 ± 0.5 (4.4–5.3) 4.8 ± 0.4 (4.4–5.3)
CC 4.4 ± 0.3 (4.2–4.9) 3.5 ± 0.9 (2.2–4.7) 3.5 ± 0.5 (3.1–4.1) 3.1 ± 1.1 (2.2–4.7)
IL 31.4 ± 2.1 (27.7–32.9) 31.0 ± 3.9 (24.6–37.8) 28.2 ± 1.8 (26.1–29.6) 27.6 ± 2.4 (25.7–31.6)
FLL 11.3 ± 0.3 (10.9–11.6) 10.9 ± 0.9 (9.5–11.9) 9.3 ± 0.1 (9.2–9.4) 10.2 ± 0.8 (9.1–11.1)
HLL 10.4 ± 0.3 (10.1–10.8) 10.0 ± 0.8 (8.36–11.1) 8.6 ± 0.6 (7.9–9.1) 9.0 ± 0.8 (7.8–9.8)
UL 16.4 ± 0.9 (16–18) 14.7 ± 1.3 (13–17) 14.7 ± 1.5 (13–16) 12.6 ± 1.7 (10–14)
LL 15.8 ± 1.1 (14–17) 15 ± 0.9 (14–16) 15.7 ± 0.6 (15–16) 13.6 ± 1.1 (12–15)
TL/
SVL
1.3 ± 0.2 (1.2–1.5) 1.2 ± 0.1 (1.1–1.4) 1.3 ± 0.1 (1.2–1.4) 1.3 ± 0.1 (1.1–1.4)
Linear measurements (in mm) and scale counts are given as mean ± standard deviation, followed by range in parentheses. M = adult male; F = adult
female; NP = national park. See text for explanation of character abbreviations.
*Included in these data are measurements from the holotype (adult female) of K. adolfifriderici (ZMB 22709).
Downloaded from https://academic.oup.com/zoolinnean/article-abstract/181/2/400/3836902
by University of Texas at El Paso user
on 28 November 2017
414 D. F. HUGHES ET AL.
© 2017 The Linnean Society of London, Zoological Journal of the Linnean Society, 2017, 181, 400–438
Differential diagnosis: A medium-sized forest chame-
leon that is distinguished from most congeners by the
absence of a rostral process in both sexes [K. asheo-
rum, K. boehmei (Lutzmann & Nečas, 2002), K. carpen-
teri, K. fischeri, K. magomberae Menegon et al. (2009),
K. matschiei (Werner, 1895), K. msuyae Menegon et al.
(2015), K. multituberculata (Nieden, 1913), K. oxyrhina
(Klaver & Böhme, 1988), K. tavetana (Steindachner,
1891), K. tenuis (Matschie, 1892), K. uluguruensis
(Loveridge, 1957), K. uthmoelleri (Müller, 1938), K. van-
heygeni Nečas, 2009, K. vosseleri (Nieden, 1913) and
K. xenorhina]. The new species can be distinguished
from K. adolfifriderici by its larger cranial crest gap
and head length, larger body size (52.8–58.7 mm vs.
47.9–54.9 mm), and more upper (16–18 vs. 10–14) and
lower labials (14–17 vs. 12–15). The new species can
be distinguished from K. tolleyae sp. nov. by the lack
of two distinctly bulging and rounded portions of the
upper casque, slightly larger head width, larger fleshy
papillae medial to rotulae on hemipenis and more
upper labials (16–18 vs. 13–17). The new species can
be distinguished from K. itombwensis sp. nov. by its
larger fore- and hind limbs, slightly larger cranial crest
gap and head length, and more conical tubercles on
the dorsal crest (8–10 vs. 6–7). The new species can be
distinguished from K. mulyai and K. excubitor by the
presence of a dorsal crest with 8–10 conical tubercles
and marked mitochondrial sequence divergence. The
new species can be distinguished from K. gyrolepis by
its smaller mean body size (55.9 vs. 67.3 mm) and cur-
rent distribution in moist Afromontane rainforest.
Genetic differentiation and variation: A summary
of pairwise sequence divergence for each molecular
marker (16S, ND2 and RAG1) among individuals of
K. rugegensis sp. nov. and other species of Kinyongia
endemic to the AR are presented in Table S2. For the
ND2 locus, p-distances among K. rugegensis sp. nov.
samples ranged from 0.1 to 0.7%.
Description of holotype: Adult female, SVL 56.6 mm
and TL 67.3 mm. Four oviductal eggs present (see
Reproduction in the following text). Casque low, slightly
raised above nape. Distinct, elevated apex on posterior
casque. Neck distinct from head. Parietal crest largely
indistinct with few enlarged and flattened tubercles in
an inconsistent pattern. Supra-orbital ridges mostly
smooth. Temporal crest comprises three enlarged
tubercles extending posteriorly from mid-eye and
ascending along posterior ridge of casque to its apex.
Nares open laterally and in a posterior orientation.
Canthal ridge consists of four slightly raised tuber-
cles, one raised higher than others near snout. Sixteen
upper and 14 lower labials are present along tip of
snout to posterior margin of orbit. No gular or ventral
Figure 6. Photographs of various individuals of Kinyongia rugegensis sp. nov. in life. (A) Adult female (gravid) lateral
view of holotype (UTEP 21485); (B) Adult male lateral view (UTEP 21481); (C) Adult male lateral view (UTEP 21482); (D)
Adult female (gravid) in aggressive posture and coloration (UTEP 21484).
Downloaded from https://academic.oup.com/zoolinnean/article-abstract/181/2/400/3836902
by University of Texas at El Paso user
on 28 November 2017
SYSTEMATICS OF ALBERTINE RIFT FOREST CHAMELEONS (CHAMAELEONIDAE: KINYONGIA) 415
© 2017 The Linnean Society of London, Zoological Journal of the Linnean Society, 2017, 181, 400–438
crests present. Nine small conical tubercles present on
anterior portion of dorsal crest, absent near mid-body.
Tail and lateral flanks smooth. Body covered in nearly
homogenous, flattened tubercles. Some larger polygo-
nal tubercles present on dorsal flanks. Patches of small
tubercles in rosette patterns on ventral flanks. Some
enlarged flattened tubercles present on outer portions
of limbs. Claws typical of Kinyongia species.
Coloration of holotype (in ethanol): Photographs of the
body and head detail of the holotype (in preservative)
are presented in Fig. 10. The background coloration is
greyish blue with some darker blotches on the flanks
and tail. Patches of lighter blues and greens are pre-
sent near the anterior portions of the body, and the
sides of the head and tail. Light yellow (almost white)
patches occur near axillary and inguinal regions and a
few places on the lateral body flanks. The soles of the
feet are yellowish-white.
Coloration of holotype (in life): A photograph of the
holotype (in life) is presented in Fig. 6A. The top of the
head is covered in brown and dark green tubercles with
black interstitium. Beginning below the temporal crest,
the head is lighter green in colour and covered in yel-
lowish-green tubercles with powder-blue interstitium.
At mid-eye, there is a dark brown lateral stripe that
connects the coloration on the canthal ridge to the tem-
poral crest. Near the tip of the snout is a pronounced
yellow coloration that fades posteriorly. The background
coloration of the body is yellowish-green with black
interstitium. The powder-blue coloration of the head
interstitium extends posteriorly on the ventral flanks
and gradually changes to black by mid-body, then blue
reappears on the posterior third of body. Tubercles on
the venter, near axillary and inguinal regions, and hid-
den parts of the limbs, are off-white with flecks of green.
The dorsal crest is adorned with darker green tubercles
than elsewhere on the body and this coloration extends
onto the tail. The posterior third of the tail is darker
brown and the greenish coloration of the tail in general
is less bright compared to the body. Differential distri-
bution of interstitial coloration (light blue or black) on
the body form broad vertical dark brown bands.
Hemipenis: Hemipenal drawings and description are
based on specimen UTEP 21481. Line drawings depict-
ing the general hemipenis morphology of K. rugegen-
sis sp. nov. are presented in sulcal and lateral views
(Fig. 7A). Hemipenes are calvate and the pedicel is less
than one-fifth of the hemipenis length. The truncus
is covered with calyces ranging in size from smaller
on the asulcal apex to larger ones near the asulcal
pedicel. Distal calyces are smaller and more hexago-
nal in shape. The sulcal lips and sulcus spermaticus
are smooth and devoid of ornamentation. The flesh
on the sulcus is highly envaginated (folded), forming
numerous sulcal ridges. Sulcal lips diverge towards
the apex and continue as a ridge that encircles the
apex. The apex is bilobed and each lobe possesses a
large, sharply denticulated rotulae. A sizeable protu-
berant fleshy papilla is positioned medially from each
rotulae.
Variation: Descriptive morphometrics of K. rugegen-
sis sp. nov. are presented in Table 4, and a summary
of mean measurements in Table 3. Chameleon photo-
graphs displaying colour variation in life are presented
in Fig. 6. Morphological proportions in paratopotypes
and paratypes are generally consistent with those in
the holotype. Males have longer tails than females
[M: 85.3 ± 6.9 (80.4–90.2 mm, n = 2); F: 66.7 ± 0.9
(65.6–67.3 mm, n = 3)] (P < 0.01), but similar body
sizes [M: 56.9 ± 2.6 (55.0–58.7 mm, n = 2); F: 55.4 ± 2.3
(52.8–56.8 mm, n = 3)] (P > 0.05). Males have overall
yellowish-brown background coloration, in contrast
to the lighter green colour of females. When agitated,
the tip of the snout, eye skin and various regions on
the flanks can be brightly coloured with yellow. One
female (UTEP 21484) in an aggressive posture and
coloration with an open mouth, showed a dark patch
laterally at mid-body, white gular and ventral regions
and bright yellow areas on the head (Fig. 6D).
Reproduction: The holotype (UTEP 21485) with SVL
56.6 mm and TL 67.3 mm collected on 23 December
2011 was gravid. This individual contained four ovi-
ductal (shelled) eggs with mean dimensions (in mm),
length 12.75 ± 0.21 (range: 12.49–12.93) and width
6.49 ± 0.25 (range: 6.31–6.84). Exact measurements of
eggs were as follows: 12.65 L × 6.84 W; 12.49 L × 6.32
W; 12.91 L × 6.49 W; 12.93 L × 6.31 W. This individ-
ual had moderate fat bodies. Another female (UTEP
21484) with SVL 52.8 mm and TL 55.6 mm collected
on 25 December 2011 was also gravid. This individual
contained four enlarged, yolked ovarian follicles with
mean dimensions (in mm), length 7.09 ± 0.19 (range:
6.83–7.25) and width 5.48 ± 0.26 (range: 5.2–5.83).
Exact follicular measurements were as follows: 7.25 L
× 5.43 W; 7.19 L × 5.2 W; 6.83 L × 5.83 W; 7.09 L ×
5.44 W. This individual possessed extensive fat bod-
ies. Conversely, a female (UTEP 21483) with SVL
58.8 mm and TL 67.1 mm collected on 22 December
2011 was not gravid, as demonstrated by the largest
ovarian follicles measuring <3 mm in diameter and
lacking evidence of yolk. This individual had minor fat
bodies.
All males had darkly pigmented testes (i.e. black
coloration), which is characteristic of all chameleon
species examined to date (Tolley & Herrel, 2013).
All collected males were sexually mature. One male
(UTEP 21481) with SVL 55.0 mm and TL 80.4 mm col-
lected on 20 December 2011 had enlarged testes. The
Downloaded from https://academic.oup.com/zoolinnean/article-abstract/181/2/400/3836902
by University of Texas at El Paso user
on 28 November 2017
416 D. F. HUGHES ET AL.
© 2017 The Linnean Society of London, Zoological Journal of the Linnean Society, 2017, 181, 400–438
right testis of this individual measured 6.89 mm in
length and 5.04 mm in width. Another male (UTEP
21482) with SVL 58.7 mm and TL 90.2 mm collected
on 23 December 2011 also had enlarged testes. The
right testis of this individual measured 6.45 mm in
length and 4.51 mm in width. Fat bodies were minor
for both of these individuals.
Diet: All five specimens examined for gut contents
had remains of arthropod prey items that could
be identified to order. The stomach of one female
(UTEP 21484) contained Hemiptera, Lepidoptera,
Hymenoptera and Coleoptera. A second female (UTEP
21485) stomach contained Diptera, Hemiptera,
Araneae and Acari. The stomach of a third female
(UTEP 21483) contained Araneae, Orthoptera and
Hemiptera. A male (UTEP 21481) stomach con-
tained Diptera and Hemiptera. Another male (UTEP
21482) stomach contained Diptera, Hemiptera,
Araneae and Psocoptera.
Distribution and natural history: Kinyongia rugegen-
sis sp. nov. is found in moist Afrotemperate montane
and sub-montane forests at an elevation range from
1660 to 2130 m. Most specimens were collected from
forest edges near and inside Kibira National Park.
This montane forest extends from southern Rwanda
(Nyungwe Forest National Park) to northern Burundi
(Kibira National Park). We speculate that this new
species is present throughout the Rugege Highlands
Figure 7. Sulcal and lateral views of the male hemipenes. (A) Kinyongia rugegensis sp. nov. (UTEP 21481) demonstrat-
ing large fleshy papillae medial to large rotulae; (B) Kinyongia tolleyae sp. nov. (UTEP 21488) illustrating small fleshy
papillae medial to large rotulae.
Downloaded from https://academic.oup.com/zoolinnean/article-abstract/181/2/400/3836902
by University of Texas at El Paso user
on 28 November 2017
SYSTEMATICS OF ALBERTINE RIFT FOREST CHAMELEONS (CHAMAELEONIDAE: KINYONGIA) 417
© 2017 The Linnean Society of London, Zoological Journal of the Linnean Society, 2017, 181, 400–438
in areas of suitable forest habitat. For example,
Hinkel (1993) recorded Bradypodion adolfi-friderici
(= Kinyongia adolfifriderici) from both Nyungwe and
Cyamudongo forests (= Nyungwe Forest National
Park) in Rwanda, and these records potentially rep-
resent this new species. Two specimens (UTEP 21482
and UTEP 21483) were collected inside a banana tree
plantation just outside the national park. One speci-
men (UTEP 21484) was collected from natural road-
side vegetation, and was found c. 2.5 m above ground
in a small tree. Two of the three females were gravid,
and both males were sexually mature. No juveniles
were detected during the search period (c. 3.5 weeks).
Behaviour and activity patterns are essentially
unknown, but likely similar to that of K. adolfifrider-
ici (Tilbury, 2010). Other lizard species collected near
the type locality included typical AR lizard fauna,
including Adolfus africanus, Chamaeleo dilepis,
Congolacerta vauereselli, Hemidactylus mabouia,
Lygodactylus cf. gutturalis, Rhampholeon boulengeri,
Trioceros ellioti, T. johnstoni, Trachylepis striata and
T. maculilabris.
Conservation: Nyungwe Forest National Park is the
largest protected area in Rwanda, and Kibira National
Park is the largest protected area in Burundi. These
contiguous parks together form one of the largest mon-
tane forest blocks in eastern Africa (Barakabuye et al.,
2007). However, despite this high level of connectivity,
similarity of threats and biodiversity importance,
these forests have been managed in near isolation to
the neighbouring protected areas (Barakabuye et al.,
2007). The highlands of these countries are renowned
for their nutrient-rich soils. As a result, the regions
are burdened with extremely dense human popula-
tions that greatly threaten the biological integrity of
the remaining forests with severe agricultural pres-
sures. Moreover, a longstanding history of armed
conflict in the region has left a legacy of irreparable
anthropogenic damage in these fragile ecosystems
(Kanyamibwa, 1998).
Etymology: The specific epithet is derived from Rugege
Highlands, the greater mountainous region where
the species was collected, with the Latin suffix –ensis
denoting a place or locality. Although the holotype
was collected from Kibira National Park in northern
Burundi, this Afromontane forest is contiguous with
Nyungwe Forest National Park in Rwanda via the
Rugege Highlands. This new species likely occurs in
suitable forested habitat across this mountain range.
The view that these neighbouring protected areas
are independent is outdated, and unfortunately, park
management in bordering countries has sustained this
position for some time (Barakabuye et al., 2007). Thus,
we felt that the taxonomy should reflect the natural
connectivity of the region and chose a broader name
accordingly.
Table 4. Descriptive morphometrics (measurements and meristic counts) for adult type specimens of Kinyongia
rugegensis sp. nov. See text for explanation of character abbreviations
UTEP 21485
Holotype female*
UTEP 21481
Paratopotype male
UTEP 21482
Paratopotype
UTEP 21483
Paratype female
UTEP 21484
Paratype female*
SVL 56.6 55.0 58.7 56.8 52.8
TL 67.3 80.4 90.2 67.1 65.6
ToL 123.9 135.4 148.9 123.8 118.4
HL 15.4 16.7 18.2 16.6 15.9
HW 7.8 8.3 8.9 8.8 7.9
HH 8.9 10.5 10.3 10.3 9.2
ML 10.2 11.4 13.0 12.0 11.6
CE 6.7 7.9 7.5 7.1 6.5
SL 5.1 5.0 6.4 5.5 5.5
ED 5.0 6.3 6.0 5.3 4.9
CC 4.3 4.2 4.9 4.5 4.4
IL 32.9 31.3 32.8 32.0 27.7
FLL 11.3 11.6 11.4 10.9 11.0
HLL 10.2 10.8 10.2 10.6 10.1
UL 16 16 18 16 16
LL 14 16 16 16 17
CTD 9 9 10 8 10
TL/SVL 1.2 1.5 1.5 1.2 1.2
*Enlarged ovarian follicles present in body cavity of specimen.
Downloaded from https://academic.oup.com/zoolinnean/article-abstract/181/2/400/3836902
by University of Texas at El Paso user
on 28 November 2017
418 D. F. HUGHES ET AL.
© 2017 The Linnean Society of London, Zoological Journal of the Linnean Society, 2017, 181, 400–438
family chamaeleoniDae gray, 1825
genus Kinyongia Tilbury, Tolley & branch, 2006
Kinyongia tolleyae sp. nov.
(figs 1, 2, 3, 4, 7b, 8, 10; Tables 3, 5)
urn:lsid:zoobank.org:act:A7494073-4318-474F-B8FC-
D4EA3692948C
Chamaeleo adolfifridericiDrewes & Vindum, 1998:
table 1, fig. 3 – Photograph in life and basic collec-
tion details.
Chamaeleo adolfifridericiVonesh, 2001: table 3 –
Record for Kibale National Park, Uganda.
Kinyongia adolfifridericiTilbury et al., 2006: table 1,
fig. 2 – Phylogenetic position.
Kinyongia adolfifridericiMenegon et al., 2009: fig. 1
– Phylogenetic position.
Kinyongia adolfifridericiBranch & Tolley, 2010:
fig. 4 – Phylogenetic position.
Kinyongia adolfifridericiTilbury, 2010: fig. 376 –
Photograph in life.
Kinyongia adolfifridericiTownsend et al., 2011: fig. 1
– Phylogenetic position.
Kinyongia adolfifridericiTolley et al., 2011: figs. 2, 3,
4 – Phylogenetic position.
Kinyongia adolfifridericiTolley et al., 2013: figs. 1,
2 – Phylogenetic position.
Kinyongia adolfifridericiGreenbaum et al., 2012a:
fig. 2, Appendix II – Phylogenetic position and
genetic distances.
Kinyongia adolfifridericiTilbury & Tolley, 2015:
fig. 4 – Phylogenetic position.
Kinyongia adolfifridericiMenegon et al., 2015: fig. 3
– Phylogenetic position.
Common name: Tolley’s forest chameleon.
Holotype: UTEP 21490 (field no. ELI 2755), adult
female, UGANDA, Western Region, Kigezi sub-region,
Kabale District, Bwindi Impenetrable National Park,
near Ruhija village, 01°254.096S 29°4636.624E,
2284 m elevation, 26 May 2014, collected at night from
natural vegetation along a roadside near Institute for
Tropical Forest Conservation (ITFC) by C. Kusamba,
M.M. Aristote and W.M. Muninga (Fig. 8E).
Paratopotypes: Same collection details as holotype, two
adult females, UTEP 21486 (field no. ELI 2754) and
UTEP 21487 [field no. ELI 2788 (28 May 2014)], col-
lected at night from forest edges c. 3 m above ground
along a road to ITFC, and one adult male, UTEP 21488
(field no. ELI 2756), collected at night with aid of stick
from c. 5 m above ground in sleeping perch of tree
behind ITFC (main office) by D.F. Hughes, K.A. Tolley,
S. Davies and A.A. Turner.
Paratype: One adult male, UTEP 21489 (field no.
ELI 2827), UGANDA, Western Region, Rwenzururu
sub-region, Kasese District, near Rwenzori Mountains
National Park, Ruboni village, 00°2058.992N
30°147.028E, 1655 m elevation, 31 May 2014, col-
lected at dusk from c. 3 m above ground in sleep-
ing perch of vegetation (secondary forest) in front
of the Ruboni Community Hotel by D.F. Hughes,
E. Greenbaum and M. Behangana.
Referred specimens: One adult female [CAS 176920 (field
no. JVV-1367)], UGANDA, Western Region, Kigezi sub-
region, Kabale District, Bwindi Impenetrable National
Park, Mubwindi Swamp, c. 120 m south of swamp, 2133
m elevation, 01°412S, 29°450E, 9 December 1990, col-
lected c. 60 cm above ground on fern by J.P. O’Brien and
J.V. Vindum. Three adult females [CAS 201593–95 (field
nos. JVV-4058–59, 4577)], UGANDA, Western Region,
Kigezi sub-region, Kabale District, Bwindi Impenetrable
National Park, ITFC near Ruhija village, 2362 m eleva-
tion, 1°247.8S, 29°4628.5E, 12 September 1996 (CAS
201593–94) and 18 October 1996 (CAS 201595), col-
lected at night c. 3 m above ground on road-cut vegeta-
tion (CAS 201593–94) and c. 2 m above ground in bush
(CAS 201595) by J.V. Vindum (CAS 201593–94), and
R.C. Drewes and J.V. Vindum (CAS 201595).
Diagnosis: Kinyongia tolleyae sp. nov. can be distin-
guished from all other Kinyongia species by the fol-
lowing combination of traits: (1) lack of rostro-nasal
ornamentation in both sexes; (2) moderate body size
(mean SVL = 56.6 mm); (3) anterior dorsal keel with
5–10 conical tubercles; (4) casque slightly elevated
above the nape; (5) two smooth, expanded areas pre-
sent on the casque that appear bilobed when viewed
from above; (6) absence of both a gular and ventral
crest; (7) 13–17 upper and 14–16 lower labials; (8) tail
length longer than SVL in both sexes; (9) parietal crest
with several slightly raised tubercles that fork towards
the snout; (10) background coloration of the body in
adult females is generally light green to yellow-green;
background coloration of the body in adult males is
generally light brown with anteriorly positioned green
patches and peach speckling near the head; (11) large
dark brown patches with white centres are present
on the lateral flanks of adult females and these lat-
eral patches are typically oriented with a larger patch
positioned anteriorly and sometimes a second smaller
patch positioned posteriorly from mid-body; (12) areas
of darker brown pigment cover the cloacal region and
extend distally onto hidden parts of the hind limbs
and tail in adult females; (13) interstitial skin between
the tubercles on the body is generally white and some-
times green for both sexes; (14) a brown stripe passes
through the middle of the eye and extends from the
canthal ridge to the temporal crest, and the eye skin
above and below the stripe is powder blue/teal, gradu-
ally dissipating dorsally and ventrally; (15) the top of
Downloaded from https://academic.oup.com/zoolinnean/article-abstract/181/2/400/3836902
by University of Texas at El Paso user
on 28 November 2017
SYSTEMATICS OF ALBERTINE RIFT FOREST CHAMELEONS (CHAMAELEONIDAE: KINYONGIA) 419
© 2017 The Linnean Society of London, Zoological Journal of the Linnean Society, 2017, 181, 400–438
the head is somewhat darker green than elsewhere;
(16) gular region and ventral portions of the body are
distinctly off-white.
Differential diagnosis: A medium-sized forest cha-
meleon that is distinguished from most other conge-
ners by the absence of a rostral process in both sexes
(K. asheorum, K. boehmei, K. carpenteri, K. fischeri,
K. magomberae, K. matschiei, K. msuyae, K. multitu-
berculata, K. oxyrhina, K. tavetana, K. tenuis, K. ulugu-
ruensis, K. uthmoelleri, K. vanheygeni, K. vosseleri and
K. xenorhina). The new species can be distinguished
from K. adolfifriderici by its larger snout length and
more upper (13–17 vs. 10–14) and lower (14–16 vs.
12–15) labials. The new species can be distinguished
from K. rugegensis sp. nov. by the presence of two dis-
tinctly expanded and smooth portions of the upper
casque (bilobed appearance), slightly smaller head
width, fewer upper labials (13–17 vs. 16–18), and
smaller fleshy papillae medial to rotulae on hemi-
penis. The new species can be distinguished from
K. itombwensis sp. nov. by its larger snout length,
slightly larger forelimbs and casque–eye distance and
generally more conical tubercles on the dorsal crest
Figure 8. Photographs of various individuals of Kinyongia tolleyae sp. nov. in life. (A) Adult male lateral view (UTEP
21489) from Rwenzori Mountains National Park; all others from Bwindi Impenetrable National Park; (B) Adult female
(gravid) lateral view (UTEP 21487); (C) Adult male lateral view (UTEP 21488); (D) Two adult females (UTEP 21486, UTEP
21490) in presence of male (not pictured); (E) Adult female displaying dark coloration (UTEP 21490); (F) Adult female lat-
eral view (UTEP 21486).
Downloaded from https://academic.oup.com/zoolinnean/article-abstract/181/2/400/3836902
by University of Texas at El Paso user
on 28 November 2017
420 D. F. HUGHES ET AL.
© 2017 The Linnean Society of London, Zoological Journal of the Linnean Society, 2017, 181, 400–438
(5–10 vs. 6–7). The new species can be distinguished
from K. mulyai and K. excubitor by the presence of a
dorsal crest with 5–10 conical tubercles and marked
mitochondrial sequence divergence. The new species
can be distinguished from K. gyrolepis by a smaller
mean body size (56.6 mm vs. 67.3 mm) and current
distribution in moist Afromontane rainforest.
Genetic differentiation and variation: Summary of
pairwise sequence divergence for each molecular
marker (16S, ND2 and RAG1) among individuals
of K. tolleyae sp. nov. and other species of Kinyongia
endemic to the AR are presented in Table S2. For the
ND2 locus, p-distances among K. tolleyae sp. nov. sam-
ples ranged from 0.0 to 1.4%.
Description of holotype: Adult female, SVL 52.9 mm
and TL 75.6 mm. Four rounded ovarian follicles pre-
sent (see Reproduction in the following text). Casque
slightly elevated above nape. Posterior apex of casque
present, overhanging nape. Two distinct expanded
areas of flattened tubercles present on top of casque,
bilobed appearance. Neck distinct from head. Parietal
crest consists of four discrete, enlarged tubercles
extending posteriorly as a ridge to apex of casque.
Supra-orbital ridges mostly smooth and one larger
conical tubercle near dorsal posterior margin of orbit
present. Temporal crest consists of three enlarged
tubercles extending posteriorly from mid-eye and
ascending posteriorly along ridge of casque to apex.
Nares open laterally and posteriorly. Canthal ridge
consists of five raised tubercles descending from eye
towards snout. Fifteen upper and 16 lower labials pre-
sent along tip of snout to posterior margin of orbit. No
gular or ventral crests present. Nine raised conical
tubercles present on anterior portion of dorsal crest,
absent near mid-body. Tail and lateral flanks smooth.
Body covered in nearly homogenous, flattened tuber-
cles. Some larger polygonal tubercles present dorsally
on flanks. Rosette patches of smaller tubercles on ven-
tral portion of body. Mostly enlarged flattened tuber-
cles present on outer portions of limbs. Claws typical
of Kinyongia species.
Coloration of holotype (in ethanol): Photographs of the
body and head detail of the holotype (in preservative)
are presented in Fig. 10. The background coloration is
various shades of blue with darker grey-blue areas cov-
ering some dorsal parts of the body and tail. The ven-
ter, beginning below the nape and extending near the
cloacal region, is a pink to off-white coloration. Patches
of lighter blues are present behind the eye, near com-
missure of the mouth, side and top of the casque, sides
of the tail and parts of the hind limbs. A large portion,
about midway on the tail, is off-white. The axillary and
inguinal regions are lighter blue-green than elsewhere
on the body. The soles of the feet are yellowish-white.
Coloration of holotype (in life): A photograph of the
holotype (in life) is presented in Fig. 8E. The follow-
ing description is based on colour photographs of the
holotype, which were taken when the animal was in
a slightly defensive display with overall darker body
colour (Fig. 8E). See Diagnosis of K. tolleyae sp. nov.
for description of more normal coloration, and see pho-
tos of other individuals in various physiological states
(Fig. 8). The top of the head is covered in dark brown
tubercles with black interstitium. The head is lighter
in colour beginning below the temporal crest to can-
thal ridge, and covered in light brown and yellowish-
green tubercles with off-white interstitium. At mid-eye,
there is a dark brown lateral stripe that connects the
coloration on the canthal ridge to the temporal crest.
The eye skin is dark brown and resembles that of the
top of head. Labial scales are heterogeneous in colour
with hues of red. The gular region is off-white, and this
coloration extends across the venter and parts of the
tail. Areas of darker brown pigment cover the cloacal
region, hidden parts of the hind limbs and part of the
tail. Area below jaw on gular is peach colour. The back-
ground coloration of the body is greenish-brown with
off-white to green interstitium. The ventral flanks are
adorned with a large dark patch of coloration, positioned
slightly anterior from mid-body. The centre of the patch
is lighter than the black edges, and almost orange in
colour. Several faint dark brown vertical bands begin
on the dorsal keel and quickly fade ventrally, not to
reach mid-body. Smaller body tubercles form rosettes,
with light green colour filling spaces between tubercles.
Tubercles near axillary and inguinal regions, and hid-
den parts of the limbs, are white with flecks of green.
The dorsal crest is ornamented with darker tubercles
than elsewhere and this pattern extends onto the tail.
The posterior third of the tail is darker green than the
rest of the tail, and faint vertical dark brown bands are
present, especially towards the distal end of the tail.
Hemipenis: Hemipenal drawings and description are
based on specimen UTEP 21488. Line drawings depict-
ing the general hemipenis morphology of K. tolleyae sp.
nov. are presented in sulcal and lateral views (Fig. 7B).
The hemipenis of this new species is very similar to
that of K. rugegensis sp. nov., except that it possesses
smaller fleshy papillae medial to each large rotulae.
See Hemipenis of K. rugegensis sp. nov. for description
of hemipenis morphology.
Variation: Descriptive morphometrics of K. tolleyae sp.
nov. are presented in Table 5 and a summary of mean
measurements in Table 3. Chameleon photographs
displaying colour variation in life are presented in
Fig. 8. Morphological proportions in paratopotypes
and paratypes are generally consistent with those
in the holotype. Males and females have similarly
Downloaded from https://academic.oup.com/zoolinnean/article-abstract/181/2/400/3836902
by University of Texas at El Paso user
on 28 November 2017
SYSTEMATICS OF ALBERTINE RIFT FOREST CHAMELEONS (CHAMAELEONIDAE: KINYONGIA) 421
© 2017 The Linnean Society of London, Zoological Journal of the Linnean Society, 2017, 181, 400–438
sized tails [M: 65.0 ± 1.1 (64.2–65.8 mm, n = 2); F:
70.7 ± 5.9 (64.0–75.6 mm, n = 7)] (P > 0.05), but males
have smaller body sizes [M: 50.1 ± 2.2 (48.5–51.6 mm,
n = 2); F: 58.4 ± 4.9 (51.5–66.2 mm, n = 7)] (P = 0.03).
Males have overall brown background coloration with
green pigmented patches anteriorly, in contrast to the
light green background coloration of females, which
are largely devoid of brown pigment. Females possess
a dark patch (sometimes two) of coloration on the lat-
eral flanks of the body with a lighter centre, whereas
males do not possess this feature. When agitated, the
lateral patches, eye skin and dorsal region of the head
became dark (Fig. 8E).
Reproduction: The female holotype (UTEP 21490) with
SVL 52.9 mm and TL 75.6 mm collected on 26 May
2014 was in the early stages of folliculogenesis. This
individual contained four slightly enlarged ovarian
follicles (completely rounded) with mean diameter (in
mm) 5.22 ± 0.22 (5.03–5.52). Exact follicular measure-
ments were as follows: 5.25 W; 5.03 W; 5.52 W; 5.09
W. This individual had moderate fat bodies. A female
paratopotype (UTEP 21487) with SVL 60.0 mm and
TL 75.4 mm collected on 28 May 2014 was gravid. This
individual contained five oviductal (shelled) eggs with
mean dimensions (in mm), length 14.05 ± 0.33 (range:
13.75–14.44) and width 7.46 ± 0.1 (range: 7.32–7.59).
Exact measurements of eggs were as follows: 14.44 L
× 7.59 W; 14.38 L × 7.44 W; 13.88 L × 7.32 W; 13.82 L ×
7.51 W; 13.75 L × 7.45 W. This individual had minor fat
bodies. A paratopotype (UTEP 21486) with a smaller
body size (SVL 51.5 mm and TL 64.3 mm) collected on
28 May 2014 was not gravid, as evidenced by the larg-
est ovarian follicles measuring <2 mm in diameter and
lacking yolk. This individual had extensive fat bodies.
Two other paratopotype females (CAS 201593 – SVL
59.5 mm and TL 64.8 mm; CAS 201594 – SVL 59.9 mm
and TL 75.6 mm) collected on 12 September 1996
were gravid. Clutch characteristics were not meas-
ured for these individuals. From a small sample, the
temporal incidence of gravidity in females at Bwindi
Impenetrable National Park seems to correspond with
the two annual peaks in precipitation for this region
(i.e. March–May and October–November), which is
a common phenomenon among chameleon species
(Tilbury, 2010). We speculate that egg production
may not occur during only one rainy period per year
or females may produce two clutches per year. More
investigation with a larger sample is warranted to
determine the seasonal reproductive cycle for females
of this new species.
All males had darkly pigmented testes (i.e. black
coloration). All collected males were sexually mature.
One male paratopotype (UTEP 21488) with SVL
48.5 mm and TL 64.2 mm collected on 26 May 2014
had enlarged testes. The right testis of this individual
measured 6.24 mm in length and 4.91 mm in width.
This individual had minor fat bodies. A male paratype
(UTEP 21489) with SVL 51.6 mm and TL 65.8 mm col-
lected on 31 May 2014 also had enlarged testes. The
right testis of this individual measured 6.65 mm in
length and 4.93 mm in width. Fat bodies for this indi-
vidual were moderate.
Diet: Three specimens examined for gut contents had
identifiable remains of arthropod prey items, one spec-
imen had an empty stomach (UTEP 21487) and one
specimen had only a bolus of unidentifiable remains
surrounded by a white mucus membrane (UTEP
21488). The stomach of one female (UTEP 21490) con-
tained Mantodea, Araneae, Hymenoptera, Diptera,
Hemiptera and Coleoptera. A second female (UTEP
21486) stomach contained Araneae and Hymenoptera.
A male (UTEP 21489) stomach contained Diptera.
Distribution and natural history: Kinyongia tolleyae
sp. nov. is found in moist Afrotemperate montane and
sub-montane forests at an elevation range from 1655
to 2362 m. Most specimens were collected from for-
est edges within Bwindi Impenetrable National Park.
Several specimens were found on sleeping perches
relatively high in the canopy (c. 5 m above ground)
and some were found lower (c. 2 m above ground).
One specimen (UTEP 21489) was collected from sec-
ondary forest on disturbed vegetation (c. 2.5 m above
ground) near the Ruboni Community Hotel just out-
side of Rwenzori Mountains National Park. The pres-
ence of this species at two disjunct mountain blocks
suggests a recent forest connection between these
areas and increases the likelihood that this species
is more widespread than currently known. For exam-
ple, Vonesh (2001) recorded Chamaeleo (= Kinyongia)
adolfifriderici from Kibale National Park in Uganda,
which is less than 50 km from Rwenzori Mountains
National Park, and thus the observation was poten-
tially this new species. We speculate that K. tolleyae
sp. nov. may also occur in other montane protected
areas with suitable forest habitat near these two sites
(e.g. forest reserves contiguous with Queen Elizabeth
National Park and Mgahinga Gorilla National Park).
Both collected males seemed sexually mature, and
four female specimens were gravid. To the best of
our knowledge, no juveniles have been detected to
date. Behaviour and activity patterns are basically
unknown, but likely similar to that of K. adolfifriderici
(Tilbury, 2010). Intersexual interactions were observed
among a few specimens before preservation. When a
male was placed in the presence of two females, male
body colour became milky white, regions on the head
greener, powder-blue eye skin became much more
striking and distinct diamond patterns suddenly
Downloaded from https://academic.oup.com/zoolinnean/article-abstract/181/2/400/3836902
by University of Texas at El Paso user
on 28 November 2017
422 D. F. HUGHES ET AL.
© 2017 The Linnean Society of London, Zoological Journal of the Linnean Society, 2017, 181, 400–438
Table 5. Descriptive morphometrics (measurements and meristic counts) for adult type specimens of Kinyongia tolleyae sp. nov. See text for explanation of
character abbreviations
UTEP 21490
Holotype
Female*
UTEP 21486
Paratopotype
Female
UTEP 21488
Paratopotype
Male
UTEP 21487
Paratopotype
Female*
UTEP 21489
Paratype Male
CAS 176920
Paratopotype
Female
CAS 201593
Paratopotype
Female*
CAS 201594
Paratopotype
Female*
CAS 201595
Paratopotype
Female
SVL 52.9 51.5 48.5 60.0 51.6 66.2 59.5 59.9 58.9
TL 75.6 64.3 64.2 75.4 65.8 75.2 64.8 75.6 64.0
ToL 128.5 115.8 112.7 135.4 117.3 141.4 124.3 135.5 122.9
HL 16.7 14.8 14.9 16.4 15.8 17.3 16.4 16.9 15.9
HW 8.2 97.7 7.3 8.6 7.7 8.2 7.8 8.1 8.0
HH 10.0 9.0 9.3 10.0 9.5 11.8 10.6 10.6 10.6
ML 12.3 10.6 10.3 12.9 11.5 13.3 13.5 13.6 12.4
CE 7.2 6.7 6.3 6.9 6.7 6.8 6.2 6.7 6.4
SL 6.0 5.1 5.3 5.9 5.7 5.8 5.7 6.0 5.6
ED 5.1 4.9 5.3 5.6 5.3 5.5 5.3 5.0 5.0
CC 4.7 4.1 3.7 4.6 4.3 3.4 2.2 2.5 2.4
IL 28.3 28.3 24.6 32.2 28.7 37.8 32.7 32.8 33.8
FLL 11.8 10.4 9.5 12.0 11.6 11.5 10.2 10.6 10.3
HLL 10.1 9.9 8.4 11.1 10.8 10.6 10.2 9.9 9.4
UL 15 16 17 15 15 13 14 14 13
LL 16 16 16 15 15 15 14 14 14
CTD 9 5 9 5 10 - - - -
TL/SVL 1.4 1.2 1.3 1.3 1.3 1.1 1.1 1.3 1.1
*Enlarged ovarian follicles present in body cavity of specimen. Eggs and CTD were not evaluated for CAS specimens.
Downloaded from https://academic.oup.com/zoolinnean/article-abstract/181/2/400/3836902
by University of Texas at El Paso user
on 28 November 2017
SYSTEMATICS OF ALBERTINE RIFT FOREST CHAMELEONS (CHAMAELEONIDAE: KINYONGIA) 423
© 2017 The Linnean Society of London, Zoological Journal of the Linnean Society, 2017, 181, 400–438
formed on the tail. Whereas female background col-
our turned a rich green, ventral portions of the body
became noticeably whiter and the lateral body patches
became marked with a brown hue at the edges and the
centre became a purer white (Fig 8D). For a detailed
list of lizard species present at the type locality, see
Drewes & Vindum (1998). Other species collected from
Rwenzori Mountains National Park comprised typical
AR lizard fauna and some endemic species, including
Adolfus jacksoni, Kinyongia carpenteri, K. xenorhina,
Leptosiaphos meleagris, Rhampholeon boulengeri,
Trioceros ellioti, T. johnstoni and T. rudis.
Conservation: Bwindi Impenetrable National Park
and Rwenzori Mountains National Park are well-
established members of the protected area network
in the AR. These areas constitute some of the few
remaining portions of intact Afromontane forests in
the Kigezi Highlands. Nevertheless, these forests face
similar anthropogenic threats to other protected areas
across the region. The current range of K. tolleyae sp.
nov. falls within the boundaries of these two protected
areas and we suspect it may be present in nearby pro-
tected areas with suitable habitat.
Etymology: The specific epithet is named in honour of
Krystal A. Tolley for her substantial contributions to
chameleon biology, with the Latin suffix –ae to denote
feminine genitive singular. To date, Krystal has partic-
ipated in the description of 12 new chameleon species,
published copious primary research articles on chame-
leons covering a remarkable breadth of subjects and
coauthored (or edited) two important books on chame-
leons (Tolley & Burger, 2007; Tolley & Herrel, 2013).
family chamaeleoniDae gray, 1825
genus Kinyongia Tilbury, Tolley & branch, 2006
Kinyongia itombwensis sp. nov.
(figs 1, 2, 3, 4, 9, 10; Tables 3, 6)
urn:lsid:zoobank.org:act:FE73D06B-2A32-44DF-
8930-C1FB6331675F
K. adolfifridericiGreenbaum et al., 2012a: fig. 2,
Appendix II – Phylogenetic placement and genetic
distances.
K. adolfifridericiTilbury & Tolley, 2015: fig. 4 –
Phylogenetic placement.
Common name: Itombwe forest chameleon.
Holotype: UTEP 20371 (field no. EBG 1605), adult
female, DRC, South Kivu Province, Mwenga Territory,
Itombwe Plateau, near Bichaka village, 03°2027.6S
28°4740.0E, 2208 m elevation, 20 June 2008, col-
lected by E. Greenbaum, C. Kusamba, M.M. Aristote
and W.M. Muninga (Fig. 9A, D).
Paratypes: One adult female, UTEP 21479 (field no. ELI
3357), DRC, South Kivu Province, Mwenga Territory,
Itombwe Plateau, Kilumbi village, 03°2556.0S
28°3434.5E, 2020 m elevation, 16 June 2015, col-
lected by M.M. Aristote (Fig. 9B–C); one adult male,
UTEP 21480 (field no. CFS 908), DRC, South Kivu
Province, Mwenga Territory, Itombwe Plateau, Miki
village, 03°2124.4S 28°4124.4E, c. 2200 m elevation,
1 October 2010, collected by M.M. Aristote.
Diagnosis: Kinyongia itombwensis sp. nov. can be
distinguished from all other Kinyongia species by
the following combination of traits: (1) lack of rostro-
nasal ornamentation in both sexes; (2) small body
size (mean SVL = 51.1 mm); (3) few conical tubercles
on dorsal crest (6–7); (4) casque almost indistinct
from nape; (5) absence of both a gular and ventral
crest; (6) 13–16 upper and 15–16 lower labials; (7)
slightly bilobed shape of the upper casque; (8) tail
length longer than SVL in both sexes; (9) parietal
crest composed of several raised tubercles forming a
semi-circle with an extension that connects posteri-
orly to apex of the casque; (10) background coloration
of the body in adult females is generally shades of
green and yellow; (11) darker brown pigment covers
the cloacal region and extends distally onto hidden
parts of the hind limbs and tail in adult females; (12)
interstitial skin between the tubercles on the body is
black, which is lighter in colour anteriorly and off-
white on the nape; (13) a brown stripe passes through
the middle of the eye, extending from the canthal
ridge to the temporal crest, and the eye skin above
and below the stripe is yellowish-green with flecks
of blue; (14) the top of the head is darker brown than
elsewhere; (15) tubercles on the casque converge to
form a weakly raised peak posteriorly; (16) dorsal
keel that is darker green-brown than elsewhere, with
incomplete vertical black bands.
Differential diagnosis: A small-sized forest chame-
leon that is distinguished from most other conge-
ners by the absence of a rostral process in both sexes
(K. asheorum, K. boehmei, K. carpenteri, K. fischeri,
K. magomberae, K. matschiei, K. msuyae, K. multitu-
berculata, K. oxyrhina, K. tavetana, K. tenuis, K. ulugu-
ruensis, K. uthmoelleri, K. vanheygeni, K. vosseleri and
K. xenorhina). The new species can be distinguished
from K. adolfifriderici by more lower labials (15–16
vs. 12–15). For differences between K. rugegensis sp.
nov. and K. tolleyae sp. nov. to K. itombwensis sp. nov.,
see their respective sections on Differential diagnosis.
The new species can be distinguished from K. mulyai
and K. excubitor by the presence of a dorsal crest
with 6–7 conical tubercles and marked mitochondrial
sequence divergence. The new species can be distin-
guished from K. gyrolepis by a smaller mean body size
Downloaded from https://academic.oup.com/zoolinnean/article-abstract/181/2/400/3836902
by University of Texas at El Paso user
on 28 November 2017
424 D. F. HUGHES ET AL.
© 2017 The Linnean Society of London, Zoological Journal of the Linnean Society, 2017, 181, 400–438
(51.1 vs. 67.3 mm) and current distribution in moist
Afromontane rainforest.
Genetic differentiation and variation: Summary of
pairwise sequence divergence for each molecular
marker (16S, ND2, and RAG1) among individuals of
K. itombwensis sp. nov. and other species of Kinyongia
endemic to the AR are presented in Table S2. For the
ND2 locus, the p-distance between two K. itombwensis
sp. nov. samples was 0.6%.
Description of holotype: Adult female, SVL 54.8 mm
and TL 63.8 mm. Casque almost indistinguish-
ably elevated above nape. Short apex on posterior
casque. Casque slightly bilobed. Neck indistinct
from head. Parietal crest consists of five enlarged
tubercles. Parietal crest tubercles in semi-circle pat-
tern at mid-casque and one distinctly larger coni-
cal tubercle present on either side. Ridge of parietal
tubercles extending to raised apex of casque. Supra-
orbital ridges smooth. Temporal crest consists of three
enlarged tubercles extending posteriorly from mid-eye
and ascending along posterior ridge of casque to apex.
Nares open laterally, in posterior orientation. Canthal
ridge consists of five raised tubercles descending from
eye towards snout and one distinctly larger conical
tubercle present anteriorly. Thirteen upper and 15
lower labials present along tip of snout to posterior
margin of orbit. No gular or ventral crests present. Six
distinctly raised conical tubercles present on anterior
portion of dorsal crest, absent far before mid-body.
Tail and lateral flanks smooth. Body covered in nearly
homogenous, flattened tubercles. Some larger polyg-
onal tubercles present dorsally on flanks. Rosette
patches of smaller tubercles present on ventral body.
Mostly enlarged flattened tubercles present on outer
portions of limbs. Claws typical of Kinyongia species.
Coloration of holotype (in ethanol): Photographs of the
body and head detail of the holotype (in preservative)
are presented in Fig. 10. The background coloration
is various shades of blue and purple with darker grey
areas on dorsal parts of the body and tail. The ven-
ter, beginning below the nape to the cloacal region, is
lighter in colour, almost pink to off-white. Patches of
lighter purple-blue are present behind the eye, near
the commissure of mouth and extend onto the gular
area. The ventral portions of the tail are off-white. The
axillary and inguinal regions are of lighter pigment
than elsewhere on the body. The soles of the feet are
yellowish-white.
Coloration of holotype (in life): Photographs of the hol-
otype (in life) are presented in Figure 9A, D. The top of
Figure 9. Photographs of two individuals of Kinyongia itombwensis sp. nov. in life. (A) Adult female lateral view (UTEP
20371); (B, C) Adult female displaying aggressive posture and coloration (UTEP 21479); (D) Adult female slightly posterior
lateral view (UTEP 20371).
Downloaded from https://academic.oup.com/zoolinnean/article-abstract/181/2/400/3836902
by University of Texas at El Paso user
on 28 November 2017
SYSTEMATICS OF ALBERTINE RIFT FOREST CHAMELEONS (CHAMAELEONIDAE: KINYONGIA) 425
© 2017 The Linnean Society of London, Zoological Journal of the Linnean Society, 2017, 181, 400–438
the head is covered in dark brown tubercles with black
interstitium. Below the temporal crest to the canthal
ridge, the head is covered in light brown and yellow
tubercles with green interstitium. At mid-eye, there
is a dark lateral stripe that connects the brown col-
oration on the canthal ridge to the temporal crest. The
skin above and below the stripe on the eye is yellow-
green with minor powder blue speckles. Labial scales
are heterogeneous in colour with mostly hues of yel-
low and brown. The gular region just below the tip of
the snout is yellow, which fades to off-white posteriorly
until entirely absent at the nape. The ventral regions
of the body are light green in colour, with shades of
white and powder blue. The background coloration
of the body is green with yellow-edged tubercles and
black interstitium. Two medium-sized grey patches
are positioned slightly anteriorly and posteriorly from
mid-body on the lateral flanks. These patches are sur-
rounded by slightly darker green tubercles. Several
dark vertical bands begin on the dorsal keel and
quickly fade ventrally, without reaching to mid-body.
Smaller body tubercles form rosettes, with light green
colour filling spaces between tubercles. Interstitial
skin on the venter is lighter than elsewhere. Tubercles
near axillary and inguinal regions, and hidden parts
of limbs, are mostly white with flecks of green. The
dorsal crest has darker green-brown tubercles than
elsewhere and this pattern extends onto the tail. The
posterior third of the tail appears darker green than
other parts of the tail, and in general, coloration of the
tail is darker than the body.
Hemipenis: Only a single male specimen was found
(UTEP 21480) and the hemipenis was not everted
upon collection in the field.
Variation: Descriptive morphometrics of K. itombwen-
sis sp. nov. are presented in Table 6 and a summary
of mean measurements in Table 3. Chameleon photo-
graphs for two individuals displaying colour variation
in life are presented in Fig. 9. Morphological propor-
tions in paratypes are generally consistent with those
in the holotype. Too few specimens have been collected
to draw reliable inferences regarding intraspecies or
intersexual variation. Also, no male photographs were
available for comparative descriptions between male
and female colour patterns in life. The following obser-
vations are based on photographs of two female speci-
mens. When agitated, the head was almost entirely
black, interstitial skin was lighter and more con-
spicuous, and large patches on the flanks were dark
brown (Fig. 9B, C). When the mouth was opened in a
defensive posture, the gular region was expanded and
displayed an off-white interstitium (Fig. 9B, C). Two
white patches, one positioned slightly anteriorly and
a second slightly posteriorly from mid-body, are pre-
sent on the lateral flanks of the female holotype, but
not present on a female paratype. Photographs of the
holotype (Fig. 9A, D) likely reflect more normal colora-
tion for the species in life, whereas photographs of a
paratype (Fig. 9B, C) are of a distressed individual in
defensive posture that is displaying aggressive colora-
tion in life.
Reproduction: Two female specimens collected on 16
June 2008 (UTEP 20371) and 20 June 2015 (UTEP
21479) were not gravid. These specimens measured
SVL 54.8 mm and 63.8 mm (UTEP 20371), and SVL
52.2 mm and 68.0 mm (UTEP 21479). The largest ovar-
ian follicles for these two individuals measured <3 mm
and the follicles lacked evidence of yolk. Fat bodies
were minor for both of these individuals. We specu-
late that the reproductive status of these females may
reflect a period with less rainfall between June and
September in the Itombwe Plateau (Jones & Harris,
2008), or that these individuals were not sexually
mature despite being of a similar body size to adults
of closely related species. More investigation with a
larger sample is necessary to determine the reproduc-
tive aspects of this new species.
A single adult male (UTEP 21480) had darkly pig-
mented testes (i.e. black coloration) and was sexu-
ally mature. This individual with SVL 52.2 mm and
TL 68.0 mm collected on 1 October 2010 had enlarged
testes. The right testis of this individual measured
6.16 mm in length and 4.17 mm in width. This indi-
vidual had minor fat bodies.
Diet: Two female specimens examined for gut contents
had empty stomachs (UTEP 20371 and UTEP 21479),
and one male specimen (UTEP 21480) had only a few
unidentifiable remains of arthropod prey items.
Distribution and natural history: Kinyongia itombwen-
sis sp. nov. is known from only three localities in the
montane forest of the Itombwe Plateau at an elevation
range from 2020 to 2208 m. The holotype was found
in the vicinity of Bichaka village in a mixed habitat
composed of primary forest and agriculture fields.
This species seems to be restricted to higher eleva-
tion montane rainforest; however, the small number
of specimens collected hindered our ability to deduce
reliable natural history information. No juveniles
were detected during multiple repeated search peri-
ods in the plateau and surrounding areas. Behaviour
and activity patterns are essentially unknown, but
likely similar to that of K. adolfifriderici (Tilbury,
2010). One male specimen (UTEP 21480) contained
a species of parasitic nematode (Rhabdias spp.) in its
lung (C. Bursey, personal communication). Other liz-
ard species collected from Itombwe comprised typical
AR lizard fauna and some endemic species, includ-
ing Congolacerta vauereselli, Holaspis cf. guentheri,
Leptosiaphos blochmanni, L. graueri, Rhampholeon
Downloaded from https://academic.oup.com/zoolinnean/article-abstract/181/2/400/3836902
by University of Texas at El Paso user
on 28 November 2017
426 D. F. HUGHES ET AL.
© 2017 The Linnean Society of London, Zoological Journal of the Linnean Society, 2017, 181, 400–438
Figure 10. Comparison of lateral views and expanded head views of the holotypes (in descending order): Kinyongia ado-
lfifriderici ZMB 22709, K. rugegensis sp. nov. UTEP 21485, K. tolleyae sp. nov. UTEP 21490 and K. itombwensis sp.
nov. UTEP 20371. Body size as snout–vent length (SVL) and tail-length (TL) are presented above each specimen. Scale
bars represent 5 mm.
Downloaded from https://academic.oup.com/zoolinnean/article-abstract/181/2/400/3836902
by University of Texas at El Paso user
on 28 November 2017
SYSTEMATICS OF ALBERTINE RIFT FOREST CHAMELEONS (CHAMAELEONIDAE: KINYONGIA) 427
© 2017 The Linnean Society of London, Zoological Journal of the Linnean Society, 2017, 181, 400–438
boulengeri, Trachylepis varia, Trioceros johnstoni and
T. schoutedeni.
Conservation: Given the extremely high level of ver-
tebrate endemism harboured in the Itombwe Plateau
and the known range of this new species as it currently
stands, it is possible that this species is endemic to
the Itombwe and Kabobo plateaus. Although gazetted
as a reserve in 2006, anthropogenic pressures in this
region are substantial and pose serious threats to the
biological integrity of Itombwe’s forest and its resident
fauna (reviewed by Greenbaum & Kusamba, 2012).
Etymology: The specific epithet is derived from the
massif, Itombwe, where this species was found, with
the Latin suffix –ensis denoting a place or locality.
DISCUSSION
phylogeneTic paTTerns anD
Taxonomic implicaTions
Results presented here are generally consistent with
the phylogenetic relationships within Kinyongia
recovered by Mariaux, Lutzmann & Stipala (2008),
Menegon et al. (2009), Tolley et al. (2011), Greenbaum
et al. (2012a), Tolley et al. (2013), Menegon et al.
(2015) and Tilbury & Tolley (2015). Also, these find-
ings support the phylogenetic positions of K. msuyae
and K. mulyai, two recently described forest chame-
leon species (Menegon et al., 2015; Tilbury & Tolley,
2015). Furthermore, our phylogenetic results sub-
stantiate the molecular relationship of K. vanheygeni
found by Menegon et al. (2015) – a species for which
genetic material was recently made available. A major
phylogenetic discrepancy among these studies is the
arrangement of species within the EAM North clade.
Specifically, the species relationships within the ‘fis-
cheri complex’ (K. boehmei, K. tavetana and K. fis-
cheri) are inconsistent across studies and currently
unresolved. A weakly supported sister relationship
between K. fischeri and K. boehmei was detected by
Tolley et al. (2011), Menegon et al. (2015) and Tilbury
& Tolley (2015). Menegon et al. (2009), Greenbaum
et al. (2012a), Tolley et al. (2013) and this study found
a sister relationship between K. boehmei and K. tavet-
ana, yet this arrangement was poorly supported in
these studies. Based solely on mtDNA, Mariaux et al.
(2008) found that these three species formed a hard
polytomy and thus phylogenetic inferences were ren-
dered ambiguous.
Another notable phylogenetic difference among stud-
ies are the unresolved species relationships within the
‘Usambara clade’ (K. multituberculata, K. matschiei
and K. vosseleri). Tolley et al. (2011) and Tilbury &
Tolley (2015) both found that this clade formed a well-
supported polytomy. Greenbaum et al. (2012a) recov-
ered a poorly supported sister relationship between
K. multituberculata and K. matschiei, with K. vosseleri
in a sister position to this clade. This study, Mariaux
et al. (2008), Menegon et al. (2009) and Menegon et al.
(2015) all recovered a sister relationship between
K. multituberculata and K. vosseleri, with K. matschiei
closely related to this clade. This arrangement was
strongly supported in this study, Mariaux et al. (2008)
and Menegon et al. (2015), yet weak support for this
sister species organization was found by Menegon
et al. (2009). Tolley et al. (2013), based on a more com-
prehensive sampling of individuals and genes, recov-
ered a strongly supported sister relationship between
K. matschiei and K. vosseleri. These relatively minor
incongruences in topology and clade support across
phylogenetic studies may reflect an underrepresen-
tation of species diversity yet to be discovered in the
EAM North clade, subtle methodological differences
for phylogenetic reconstructions, or accelerated evo-
lutionary rates that influenced long-branch attraction
and yielded non-monophyly.
The formal descriptions of these new Kinyongia
species from the AR clarify previous assertions by
Greenbaum et al. (2012a) and Tilbury & Tolley (2015)
that the taxonomy of AR forest chameleons did not
reflect the true diversity in this region. These two pre-
vious studies – which were focused on AR Kinyongia
– included genetic data for the newly described K.
Table 6. Descriptive morphometrics (measurements and
meristic counts) for adult type specimens of Kinyongia
itombwensis sp. nov. See text for explanation of charac-
ter abbreviations
UTEP 20371
Holotype
Female
UTEP 21479
Paratype
Female
UTEP 21480
Paratype
Male
SVL 54.8 46.1 52.2
TL 63.8 64.5 68.0
ToL 118.6 110.6 120.2
HL 15.9 14.4 14.9
HW 8.5 7.0 7.7
HH 9.6 8.9 8.7
ML 12.2 9.8 10.7
CE 5.8 6.2 6.4
SL 5.4 4.7 5.0
ED 5.2 4.4 5.3
CC 3.1 3.4 4.1
IL 29.6 26.1 28.8
FLL 9.2 9.4 9.4
HLL 8.7 7.9 9.1
UL 13 16 15
LL 15 16 16
CTD 6 7 7
TL/SVL 1.4 1.3 1.2
Downloaded from https://academic.oup.com/zoolinnean/article-abstract/181/2/400/3836902
by University of Texas at El Paso user
on 28 November 2017
428 D. F. HUGHES ET AL.
© 2017 The Linnean Society of London, Zoological Journal of the Linnean Society, 2017, 181, 400–438
tolleyae sp. nov. (CAS 201593–94) and K. itombwensis
sp. nov. (UTEP 20371), and both studies recovered two
distinct lineages of K. cf. adolfifriderici. Older stud-
ies that predate the availability of genetic material
for K. itombwensis sp. nov. provided by Greenbaum et
al. (2012a) (e.g. Menegon et al., 2009; Branch & Tolley,
2010; Townsend et al., 2011; Tolley et al., 2011) used
K. tolleyae sp. nov. samples (CAS 201593–94) as repre-
sentatives of K. adolfifriderici. Our study provides the
first inclusion of putative topotypic (Ituri rainforest,
DRC) material for K. adolfifriderici, and because this
recently procured population was genetically distinct
from all other K. cf. adolfifriderici samples, we are
confident that the three populations we describe rep-
resent new species. Furthermore, the phylogeographic
patterns we recovered for these cryptic species were
indicative of lineage formation in isolation and we
speculate that previously published observations of ‘K.
adolfifriderici’ from remote forest localities in the AR
represent either additional undescribed lineages or
one of the new species described herein [e.g. Virunga
National Park (de Witte, 1941), Kahuzi-Biega National
Park (Pupin et al., 2012) and several other AR forest
localities (de Witte, 1965; Spawls et al., 2002; Tilbury,
2010)]. The recent descriptions of several new cryptic
reptile species from the AR Biodiversity Hotspot (e.g.
Congolacerta asukului Greenbaum et al., 2011; K.
gyrolepis; Cordylus marunguensis Greenbaum et al.,
2012b; Boaedon radfordi Greenbaum et al., 2015; K.
mulyai and Rhampholeon hattinghi Tilbury & Tolley,
2015), including these new chameleons, suggest that
our understanding of the region’s true diversity is far
from complete. Moreover, a recent conservation assess-
ment of Africa’s reptilian fauna found that Central
Africa is one of the three most under-sampled regions
on the continent (Tolley et al., 2016). Considering
the immense biodiversity already known from the
AR (e.g. Plumptre et al., 2007), we believe that the
biological discovery of cryptic taxa in this region is
still in its infancy (see Bickford et al., 2007), a situa-
tion that is exciting for taxonomists, but problematic
for the conservation of Central Africa’s herpetofauna
(Greenbaum, In press).
DaTing esTimaTes anD
hisTorical biogeography
The estimated divergence dates we recovered for the
genus Kinyongia were more ancient than those found
by Tolley et al. (2011). Our lineage diversification
dates more closely resembled analyses by Townsend
et al. (2011) and Tolley et al. (2013). Townsend et al.
(2011) included only four species of Kinyongia, and
in turn, this greatly hampered our ability to draw
meaningful comparisons regarding diversification
dates for the genus. The younger dates proposed by
Tolley et al. (2011) may be a result of methodological
differences for calibration priors and outgroup taxa
between studies. Tolley et al. (2013), Townsend et al.
(2011) and this study all used multiple outgroup taxa,
several fossil calibrations outside of Chamaeleonidae
and secondary calibrations within chameleons. In con-
trast, Tolley et al. (2011) used fossil calibrations within
Chamaeleonidae, secondary internal node splits based
on molecular dating analyses, and much fewer out-
group taxa. An alternative dating scenario was consid-
ered by Tolley et al. (2011) that included more outgroup
taxa (sister group Agamidae) and produced much older
diversification dates (see Appendix S2 in Tolley et al.,
2011). However, the 95% HPD intervals for the latter
analysis were suspiciously inflated and thus dating
estimates were discredited as artefacts. Given incon-
sistencies for divergence dating with molecular rates
and/or narrow taxonomic scope (Sauquet, 2013), it
seems reasonable that studies incorporate sufficient
outgroup taxa and calibration priors appropriate for
the specific study system.
The divergence dates we recovered for Kinyongia
were nearly identical to those proposed by Tolley
et al. (2013), which included a much more compre-
hensive sampling of species and genes. For example,
Tolley et al. (2013) and this study both found that the
initial divergence of the three major clades within
the genus occurred in the Eocene, whereas Tolley et
al. (2011) found these dates largely occurred in the
Oligocene. For Kinyongia species endemic to the AR,
comparisons to previous studies were rendered more
difficult because several new species have been dis-
covered more recently (e.g. Greenbaum et al., 2012a;
Tilbury & Tolley, 2015; this study). Nevertheless, we
found that the sister species K. xenorhina and K. car-
penteri, both endemic to the Rwenzori Mountains,
evolved relatively recently with a divergence date in
the late Miocene (c. 7 Mya). Tolley et al. (2011) and
Tolley et al. (2013) both found the split between these
sister species to have occurred around the same time
(c. 6 Mya). Although our estimated mean divergence
dates and those of Tolley et al. (2013) were gener-
ally more ancient than dates proposed by Tolley et
al. (2011), there was a great deal of overlap for the
95% HPD of divergences. As a result, we are confi-
dent that our dates are representative of the broad
biogeographic history of this group. For example, we
found a split between the AR and Kenya Highlands
species K. excubitor dated in the late Oligocene (c. 23
Mya), as did Tolley et al. (2013) (c. 22 Mya). However,
Tolley et al. (2011) found this split to have occurred
much earlier, in the mid-Miocene (c. 17 Mya). The
95% HPD for this divergence recovered by Tolley et
al. (2011) ranged from c. 6 to 28 Mya, and this large
interval overlapped with our findings (c. 17–29 Mya)
and Tolley et al. (2013) (c. 16–29 Mya). Although
Downloaded from https://academic.oup.com/zoolinnean/article-abstract/181/2/400/3836902
by University of Texas at El Paso user
on 28 November 2017