Post on 21-May-2020
The ecological and conservation significance of camera-trap bycatch data:
a case study of two reclusive, forest-specialised bird species from the
Indian Ocean Coastal Belt Forest
1
D.A. Ehlers Smith, Y.C. Ehlers Smith & C.T. Downs
University of KwaZulu-Natal
Smithd1@ukzn.ac.za
Joint SANBI BIMP & FBIP Forum, Cape St. Francis
14-08-2018
Introduction Methods Results Discussion
Bycatch data: Species recorded that were not
the initial target of the survey
Eligibility dependent upon certain factors: • Camera-trap placement
• Original survey design & assumptions
- proposed research question
- applicability to bycatch species
• Bycatch sample size
- Replicability, robustness of analyses
Introduction Methods Results Discussion
Case study: Two reclusive,
terrestrial forest birds
Survey design: Mammalian community
Assumptions: • “Closed” survey
• Generalised for a community
• Replicable & long-term
• Explanatory covariates
appropriate & complementary
Introduction Methods Results Discussion
Population Increase
Increased Agricultural
Outputs
Increase in housing required
Conversion of natural habitats
Increased threat to
biodiversity
Increased conflict
6
• Indigenous Forest transformation = significant
changes in biodiversity 1
• Anthropogenic fragmentation of forested landscapes
is not a random process 2
- Commercial transformation
- Agriculture/Forestry
- Urbanisation
• Fragmentation =
- changes in diversity and abundance of specialised
structures, resources and concomitant niches
- changes in species richness and abundance,
community assemblages and decline in specialist
species 3
- decrease in connectivity & gene flow
- increase in “edge” habitats & effects
- increase in alien & invasive species
1 Watson et al. 2014; 2 Baldi et al. 2006; 3 Slade et al. 2013
www.adelaide.edu.au
www.desdemonadespar.net
Introduction Methods Results Discussion
Introduction Methods Results Discussion
Scientific Name: Aplopelia larvatus / Columba
larvatus (Temmink, 1809)
Common Names: Lemon Dove, Cinnamon Dove
Scientific Name: Geokichla guttata / Zoothera
guttata (Vigors, 1831)
Common Names: Spotted Ground-thrush,
Natal Thrush
Both species may be problematic to record using
“traditional” (point-count) survey techniques
• The Spotted Ground Thrush is a globally
endangered forest specialist
• Distribution and status of both species
poorly understood in the critically
endangered Indian Ocean Coastal Belt
Forest (IOCBF), east coast of southern Africa
• Historic/paleo-climatic changes
• Biogeographic forces
• Anthropogenic impacts
• > 70% of original IOCBF has been lost
• = Fragmented forest remnants within a
mosaic of different land uses.
Introduction Methods Results Discussion
Introduction Methods Results Discussion
Main Objectives: Spotted Ground Thrush
• Establish non-breeding season habitat-
patch occupancy in southern KZN
• Examine habitat requirements and
response to fragmentation effects and
surrounding land use
Main Objectives: Lemon Dove
• Establish site occupancy
• Examine habitat requirements and
tolerance to disturbance
• Establish seasonal differences
Introduction Methods Results Discussion • June 2014 – May 2016
• Camera traps deployed over 21 survey days (24 hrs, 30 s trigger)
• Across 3 different habitat types: - Coastal Scarp Forest - Coastal LL Forest - Coastal Dense Bush
• 4 management regimes: - Nature Reserves - Residential Areas - Farms - Eco-Estates
11 1Bibby et al. (2000); 2 GeoTerra Image (2014)
www.illustrationsource.com
20 m fixed radius
Micro-habitat
characteristics1
1 km buffer
Land-cover
classifications2
Introduction Methods Results Discussion
400 m x
400m
12
Calculate patch
size & isolation
distances
between forest
patches
Mainland
Scarp
Mainland
Coastal
Introduction Methods Results Discussion
Single Season Occupancy Model:
• Site occupancy Ψ – probability that a site is occupied
• Detection probability P – probability of detecting the species, given the species is present at a site
• Each sampling occasion is treated as a temporal repeat of the survey • Presence/absence matrix: naïve
occupancy > 0.2 • Patch occupancy (SGT) • Site occupancy (LD) • AIC model selection
- ΔAIC < 2 considered equivalent
14 1 MacKenzie et al (2006)
Introduction Methods Results Discussion
15
SGT: Cryptic - 42/275 camera sites & 23/82 camera patches
LD: 76/250 camera sites in spring-summer & 70/250 camera sites in autumn-winter
Introduction Methods Results Discussion
16
Spotted Ground Thrush Ψ
0.39 ± 0.09
P
0.11 ± 0.03
Introduction Methods Results Discussion
y = 0.0007x + 0.294
R² = 0.3188
0
0,2
0,4
0,6
0,8
1
0 500 1000 1500
Ψ
Patch size
y = -0.0001x + 0.5317
R² = 0.2779
0
0,2
0,4
0,6
0,8
1
0 2000 4000 6000
Ψ
Isolation distance Patch size Isolation distance
TOP MODEL: Ψ(PS+iso+BG+SD6-10+SWI)
Covariates that influenced occupancy positively
• Ψ: Patch size (ß 28.32 ± 11.91; ωi = 1)
• Ψ: % bare ground (ß 17.72 ± 9.91; ωi = 0.995)
• Ψ: Urban Forest mosaic (ß 0.92 ± 0.72; ωi = 0.374)
Covariates that influenced occupancy negatively
•Ψ: Isolation distance (ß -2.12 ± 1.04; ωi = 1)
•Ψ: Stem density of trees 6-10 m (ß -1.18 ± 0.80; ωi = 0.989)
•Ψ: SWI (ß -1.21 ± 0.85; ωi = 0.310)
•Ψ: Coastal Dense Bush (ß -1.14 ± 0.89; ωi = 0.128)
17
Lemon Dove
Ψ 0.39 ± 0.08
P
0.26 ± 0.05
Introduction Methods Results Discussion
• TOP MODEL: Ψ HC+SD6+W0+SWI (Autumn-winter)
• Ψ: Herbaceous cover (ß 28.32 ± 11.91; ωi = 1)
• Ψ: Stem density 6-10 m trees (ß 0.40 ± 0.17; ωi = 0.9943)
• Ψ: % woody cover (ß -0.27 ± 0.17; ωi = 0.9577)
• Ψ: SWI (ß 0.16 ± 0.15; ωi = 0.2846)
Ψ 0.37 ± 0.08
P
0.25 ± 0.04
Autumn-winter
(Mar-Aug)
Spring-summer
(Sep-Feb)
Autumn-winter
(Mar-Aug)
• TOP MODEL: Ψ LL+SD16+BG+GC+PSR (Spring-summer)
• Ψ: % bare ground (ß 0.28 ± 0.21; ωi = 0.9957)
• Ψ: % grass cover(ß 0.59 ± 0.40; ωi = 0.9953)
• Ψ: Stem density 16-20 m trees (ß 0.40 ± 0.26; ωi = 0.9282)
• Ψ: % leaf litter (ß 0.54 ± 0.35; ωi = 0.7297)
• Ψ: PSR (ß 0.17 ± 0.11; ωi = 0.6937)
Spring-summer
(Sep-Feb)
• Preference for tall trees with open
understory implies dependence on
Indigenous Forest (both)
• Preference for complex habitat structures
and high plant diversity, and leaf litter/
bare ground in breeding season (LD)
• Avoidance of small, isolated patches in
proximity to agriculture implies negatively
affected by fragmentation (SGT)
• The success of this study supports the use
of camera traps for surveying reclusive,
terrestrial forest birds
• The robust conclusions highlight the
application and importance of bycatch
data for conservation and ecology
www.adelaide.edu.au
Introduction Methods Results Discussion
19
• Bycatch data may provide critical insights into
the behaviour or response of biodiversity to
ecological conditions or anthropogenic pressure
• By examining the context of bycatch data,
camera-trap surveys offer a POTENTIAL
cornucopia of opportunities
• If bycatch data do not represent your research
interests, time or budget COLLABORATE!!
Overall Research Implications
20
• Land owners & conservancies
• Ezemvelo KZN Wildlife Scientific Services
• UKZN Gay Langmuir Fieldwork grant
• UKZN CAES grant
• Hans Hoheisen Charitable Trust
• Whitley Wildlife Conservation Trust
• National Research Foundation Doctoral Scholarship
• Claude Leon Post-Doctoral Research Fellowship
Acknowledgements