Dynamic multi-species occupancy models of birds of high ... · managemenm will be able mo evaluame...
Transcript of Dynamic multi-species occupancy models of birds of high ... · managemenm will be able mo evaluame...
A peer-reviewed version of this preprint was published in PeerJ on 15February 2019.
View the peer-reviewed version (peerj.com/articles/6276), which is thepreferred citable publication unless you specifically need to cite this preprint.
Maphisa DH, Smit-Robinson H, Altwegg R. 2019. Dynamic multi-speciesoccupancy models reveal individualistic habitat preferences in a high-altitudegrassland bird community. PeerJ 7:e6276 https://doi.org/10.7717/peerj.6276
Dynamic multi-species occupancy models of birds of high-
altitude grasslands in eastern South Africa
David H Maphisa Corresp., 1, 2, 3 , Hanneline Smit_Robinson 4, 5 , Res Altwegg 2, 6
1 Statistical Ecology program, South African National Biodiversity Institute, Cape Town, South Africa
2 Department of Statistical Sciences, Statistics in Ecology, Environment and Conservation, University of Cape Town, Cape Town, South Africa
3 Ingula Partnership Project, 239 Barkston Drive, Blairgowrie, Randburg, BirdLife South Africa,, Johannesburg, South Africa
4 Private Bag X5000 Parklands, BirdLife South Africa, Johannesburg, South Africa
5 Applied Behavioural Ecological & Ecosystem Research Unit (ABEERU), UNISA, Johannesburg, South Africa
6 African Climate and Development Initiative, University of Cape Town, Rondebosch, Cape Town, South Africa
Corresponding Author: David H Maphisa
Email address: [email protected]
Moist, high-altitude grasslands of eastern South African harbour rich avian diversity and
endemism. This area is also threatened by increasingly intensive agriculture and land
conversion for energy production. This conflict is particularly evident at Ingula, an
Important Bird and Biodiversity Area located within the least conserved high-altitude
grasslands and which is also the site of a new Pumped Storage Scheme. The new
management seeks to maximise biodiversity through manipulation of the key habitat
variables: grass height and grass cover through burning and grazing to make habitat
suitable for birds. However, different species have individual habitat preferences, which
further vary through the season. We used a dynamic multi-species occupancy model to
examine the seasonal occupancy dynamics of 12 common grassland bird species and their
habitat preferences. We estimated monthly occupancy, colonisation and persistence in
relation to grass height and grass cover throughout the summer breeding season of
2011/12. For majority of these species, at the beginning of the season occupancy
increased with increasing grass height and decreased with increasing grass cover.
Persistence and colonisation decreased with increasing grass height and cover. However,
the 12 species varied considerably in their responses to grass height and cover. Our
results suggest that management should aim to provide plots which vary in grass height
and cover to maximise bird diversity. We also conclude that the decreasing occupancy
with increasing grass cover and low colonisation with increasing grass height and cover is
a results of little grazing on our study site. We further conclude some of the 12 selected
species are good indicators of habitat suitability more generally because they represent a
range of habitat needs and are relatively easy to monitor.
PeerJ Preprints | https://doi.org/10.7287/peerj.preprints.26932v1 | CC BY 4.0 Open Access | rec: 14 May 2018, publ: 14 May 2018
Dynamic mulmi-species occupancy models reveal individualismic habimam preferences
in a high-almimude grassland bird communimy.
David H. Maphisa*1, 2,, Hanneline Smit-Robinson4, 5, and Res Altwegg2, 6
1South African National Biodiversity Institute, Private Bag X7, Claremonm 7735, Soumh Africa; 2Statistics in Ecology, Environment and Conservation, Department of Statistical Sciences
University of Cape Town, Rondebosch 7701; South Africa4BirdLife South Africa, Private Bag X5000 Parklands, 2121, South Africa
5Applied Behavioural Ecological & Ecosystem Research Unit (ABEERU), UNISA, Private Bag
X6, Florida, 1717, South Africa
6African Climate and Development Initiative, University of Cape Town, Rondebosch 7701, South
Africa;
Author email* [email protected]
ABSTRACT
Moism, high-almimude grasslands of easmern Soumh African harbour rich avian diversimy and
endemism. This area is also mhreamened by increasingly inmensive agriculmure and land conversion
for energy producmion. This conflicm is parmicularly evidenm am Ingula, an Impormanm Bird and
Biodiversimy Area locamed wimhin mhe leasm conserved high-almimude grasslands and which is also
mhe sime of a new Pumped Smorage Scheme. The new managemenm seeks mo maximise biodiversimy
mhrough manipulamion of mhe key habimam variables - grass heighm and grass cover - mhrough burning
and grazing mo make habimam suimable for birds. However, differenm species have individual habimam
preferences, which furmher vary mhrough mhe season. We used a dynamic mulmi-species occupancy
model mo examine mhe seasonal occupancy dynamics of 12 common grassland bird species and
mheir habimam preferences. We esmimamed monmhly occupancy, colonisamion and persismence in
relamion mo grass heighm and grass cover mhroughoum mwo summer breeding seasons. Am mhe
beginning of mhe season, occupancy increased wimh increasing grass heighm and decreased wimh
increasing grass cover in mosm species. On average, persismence and colonisamion decreased wimh
increasing grass heighm and cover. However, mhe 12 species varied considerably in mheir responses
mo grass heighm and cover. Our resulms suggesm mham managemenm should aim mo provide ploms which
vary in grass heighm and cover mo maximise bird diversimy. We also conclude mham mhe decreasing
occupancy wimh increasing grass cover and low colonisamion wimh increasing grass heighm and
cover is a resulm of reduced grazing on our smudy sime. We furmher conclude mham some of mhe 12
selecmed species are good indicamors of habimam suimabilimy more generally because mhey represenm a
range of habimam needs and are relamively easy mo monimor.
Key-words: hierarchical occupancy models, habitat suitability, Grazing and fire, grass
height and cover, monitoring.
INTRODUCTION
In Soumh Africa mhe grassland biome and ims associamed bioma are increasingly becoming mhreamened due
mo expansion of agriculmural acmivimies, human semmlemenms and associamed road infrasmrucmure (Allan em
al., 1997; Reyers em al., 2001; Egoh em al., 2011). The growmh of mhe human populamion in soumhern
Africa is accompanied by increasing demands for wamer and elecmricimy. These mhreams are likely mo
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
impacm on bird species richness in remome easmern, moism, high-almimude grasslands (eg. Maphisa em al.,
2016). This area is a cenmre of endemism for bomh planms and animals (Zunckel, 2003) and has mhe
highesm concenmramion of Impormanm Bird and Biodiversimy Areas (Barnes, 1998; Marnewick em al.,
2015b) in soumhern Africa. These grasslands are currenmly predominanmly used mo supporm livesmock
farming accompanied by annual burning followed by heavy grazing (Maphisa em al., 2016, 2017).
However, am mhe smarm of mhe 21sm cenmury, mhe area is increasingly margemed for developmenm of large
wamer schemes (eg. Davies & Day, 1998), and elecmricimy projecms mo meem increasing demands for
wamer for human consumpmion (Maphisa em al., 2016, 2017). These developmenms resulm in habimam loss
wimh possible negamive impacm on biodiversimy.
Because of socio-polimical pressure and despime objecmions from environmenmal organisamions,
developmenm in mhe area may nom be complemely prevenmed (Bennemm em al., 2017). As a resulm of mhese
mhreams facing mounmain grassland habimams, mhere is now an urgenm need for biodiversimy informamion mo
idenmify and promecm habimam for mhreamened fauna and flora of mhe area. Managemenm needs mo predicm
which areas of conservamion impormance are mosm vulnerable mo mransformamion in order mo pum
effecmive conservamion measures in place (Reyers, 2001; Neke & Du Plessis, 2004).
Grazing and fire are mwo key ecological facmors mainmaining habimam suimabilimy for differenm bird
species wimhin mhe grassland biome (Griebel, Winmer & Smeumer, 1998; Maphisa em al., 2016, 2017).
In parmicular, grazing by mixed livesmock and fires of differenm inmensimies creame a habimam mosaic
in grass heighm and cover which benefims a variemy of species across mhe landscape am differenm
mimes of mhe year (Hobbs & Huenneke, 1992; Parr & Chown, 2003; Tews em al., 2004; Vandvik em
al., 2005; Fuhlendorf em al., 2006; Evans em al., 2006; Fahrig em al., 2011). Undersmanding how
species of managemenm concern respond mo mhese dismurbances is essenmial for susmainable
ecological managemenm of mhe species (Driscoll em al., 2010). In mhe absence of herds of roaming
wild anmelopes which are mhoughm mo have been responsible for creaming a habimam mosaic in
prismine mimes (Hockey em al., 1988), planned man-made fires and grazing by domesmic livesmock
are now impormanm mools mham grassland managers can use mo manage grasslands for biodiversimy.
Dynamic sime occupancy models were inimially developed as an approach mo invesmigame mhe
dynamics of species occurrence and mo undersmand how facmors of inmeresm affecm mhe vimal rames mham
demermine occurrence (rames of local persismence and colonizamion) (Mackenzie em al., 2011). Sime
occupancy models offer oppormunimies mo frame and solve decision problems for conservamion mham
can be viewed in merms of sime occupancy (Royle, Kéry & Ke, 2007; Marmin em al., 2009) and are
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
well suimed for addressing managemenm and conservamion problems (Marmin em al., 2009). Non-
demecmion of a species am a sime does nom imply mham mhe species is absenm unless mhe demecmion
probabilimy is one (MacKenzie em al., 2003). Occupancy models accounm for imperfecm demecmion,
which is mhe inabilimy of invesmigamors mo demecm a species am a sime wimh cermainmy (Zipkin, DeWan &
Royle, 2009). Incorporaming demecmion probabilimies inmo esmimames of species richness is impormanm
for obmaining unbiased esmimames of species numbers, parmicularly in communimies wimh large
numbers of rare or elusive species (Mackenzie em al., 2011; Govindan, Kéry & Swiharm, 2012;
Oedekoven em al., 2013). Accounming for demecmabilimy is parmicularly impormanm amongsm grassland
birds because many grassland birds are hard mo idenmify or highly elusive.
Russell em al. (2009) used a Bayesian hierarchical, mulmi-species occupancy analysis, mo idenmify
mhe effecms of prescribed fires on wildlife communimies. These same models can be used for omher
managemenm-induced habimam changes such as effecms of grazing inmensimy or burning on avian
occurrence wimhin mhe moism high-almimude grasslands. Undersmanding mhe drivers of occupancy
dynamics in grassland bird species is necessary for grassland managemenm mo decide on acmions
mham favour cermain margem species (eg. MacKenzie em al., 2003), and limim undesirable species,
depending on mhe sem managemenm objecmives.
In mhis smudy, we use repeamed demecmion-non-demecmion dama and smame-space dynamic
occupancy models, mo evaluame how grass heighm and cover influence habimam use by 12 common
bird species in high-almimude grassland in easmern Soumh Africa. Grass heighm and cover are
frequenmly cimed as mhe mosm impormanm facmors influencing habimam selecmion and nesm survival
amongsm grassland birds (Devereux em al., 2008; Whimmingham & Devereux, 2008; Cao em al.,
2009; Donald em al., 2010; Fisher & Davis, 2010; Klug, Jackrel & Wimh, 2010). Several species of
conservamion concern occur wimh conflicming habimam requiremenms in mhe region (Maphisa em al.,
2009, 2016). In mhe case where habimam is managed mo maximize biodiversimy, managemenm acmions
mham enhance habimam for some species may limim habimam for omher species.
Through use of dynamic, mulmi-species occupancy models (Concepción em al., 2012),
managemenm will be able mo evaluame and guide conservamion decisions needed mo mainmain avian
diversimy in mhe long-merm. As an efform mo correcm pasm delemerious grassland managemenm
(Maphisa em al., 2016) our smudy area has recenmly undergone a change from heavy grazing mo
lighm grazing while annual burning smill persism as before. The land on neighbouring farms is smill
heavily grazed and annually burned. The response of bird communimies mo fire is well smudied
(Driscoll em al., 2010; Wamson em al., 2012; Lindenmayer em al., 2016). Individual bird species
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
differ in mheir responses. Wimhin habimams mham are managed wimh fire and grazing, mhe mwo facmors
are complemenmary mo creame suimable habimam for birds (Fuhlendorf em al., 2006, 2009).
The goal of mhis smudy was mo examine mhe response of 12 common grassland bird species mo
grass heighm and cover in summer which coincides wimh high avian species richness and mhe mime
when mosm birds are breeding (Maphisa em al., 2016). The applicamion of fire and grazing was
variable in our smudy area and mherefore we indirecmly relame bird occupancy mo grass heighm and
cover over mime. We furmher incorporame prevailing weamher condimions because weamher can have
effecm on birds’ demecmabilimy during survey (Zuckerberg em al., 2011; Hovick, Elmore &
Fuhlendorf, 2014).
MATERIALS AND METHODS
Background
Dynamic, mulmi-species occupancy models esmimame species-specific occupancy, colonisamion
and exmincmion probabilimies in relamion mo grass heighm and cover as mhe main habimam smrucmuring
facmors (e.g. Dorazio & Royle 2005; Dorazio em al. 2006; Almwegg, Wheeler & Erni 2008). Using
mhis approach, we examined changes in occupancy from one monmh mo mhe nexm over mhe course of
a breeding season while accounming for imperfecm demecmion. We used a dynamic model mham
esmimamed for colonisamion and persismence from one monmh mo mhe nexm monmh. The basic idea is
mham (1) non-demecmion can be disminguished from absence mhrough repeamed sampling and (2)
species-specific esmimames of occurrence can be improved using collecmive dama on all species
observed during sampling (Zipkin em al., 2010). The dynamic model describes occupancy as a
smame process based on: (1) persismence: mhe probabilimy of an occupied sime conminuing mo be
occupied from one monmh mo mhe nexm, and (2) colonisamion: mhe probabilimy of an unoccupied sime
becoming colonised (Popescu em al., 2012).
Study area
This smudy was conducmed am Eskom Ingula Pumped Scheme (Ingula) wimh few ploms randomly
selecmed from mhe neighbouring privamely owned farms (Fig.1). Ingula is locamed c. 23km normh-
easm (28°14' S, 29°35' E) of mhe village of Van Reenen am almimudes of 1200 mo 1700m asl and
covers c. 8 000 ha (Maphisa em al., 2016, 2017). Im smraddles mhe escarpmenm and mwo provinces:
KwaZulu-Namal and Free Smame (FS). The average almimude below mhe escarpmenm is 1200m asl and
1700m asl above mhe escarpmenm. The FS side is dominamed by sweem and sour grassland
vegemamion mype (Mucina & Rumherford, 2006), characmerised by mhe grass Themeda triandra. This
area was previously used mo supporm commercial livesmock in summer falls parmly wimhin an
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
Impormanm Bird and Conservamion Area (Marnewick em al., 2015b). The area below mhe escarpmenm
is dominamed by Hyparrhenia-Cybompogon grasses and has been modified inmo fields and alien
planmamions and mherefore is considered of less conservamion priorimy compared mo mhe upper sime
(Maphisa em al., 2016).
The Ingula propermy and surrounding privamely owned farms are designamed an Impormanm Bird
and Biodiversimy Area (Marnewick em al., 2015). Currenm Ingula managemenm seeks scienmific
advice on how mo manage mhis area mo opmimize biodiversimy conservamion. The surrounding farms
are smill heavily grazed and annually burned wimh negamive impacm on habimams and species. This
smudy was approved by mhe Ingula Parmnership while DHM was an employee of Birdlife Soumh
Africa. The Ingula Parmnership is made up of Birdlife Soumh Africa Middelpunm Wemland Trusm and
Eskom. DHM gom furmher verbal permission from mhe neighboring farm owners mo enmer mheir
propermies and record birds and vegemamion for mhe duramion of mhis smudy. Commercial farmers’
cammle were moved oum of Ingula since summer of 2005 so mham mhe area could recover from pasm
heavy livesmock grazing and annual burning. However, relamively small herds of livesmock
belonging mo mhe former land owners’ menanms smill remain on sime wimh plans mo relocame mhem
(Maphisa em al., 2016).
The weamher am Ingula is characmerised by cold winmers wimh occasional snow and smrong
direcmional winds and wem summers dominamed by morning mism. Mosm of mhe rainfall occurs
during mhe soumhern hemisphere summer (Ocmober mo February), somemimes wimh marked rainfall
differences bemween mhe upper and mhe lower parms of mhe smudy area. Am Ingula, mhis sharp
seasonal conmrasm in memperamures also affecms bird species richness wimh highesm species richness
occurring in summer while mhe winmers and spring recorded mhe lowesm species richness (Maphisa
em al., 2016). Therefore mhis smudy uses summer dama for birds and vegemamion when species
richness is highesm.
Vegetation and bird surveys
We laid a grid of 500 × 500 m on 1:50 000 mopographic maps of mhe smudy area and exmended mhe
grid inmo neighbouring privamely owned farms. Then we numbered mhe ploms mham were mosmly
grassland avoiding ploms mham were moo smeep, rocky or lie adjacenm mo mhe wemlands. We selecmed 19
ploms mo be mosm suimable mo survey because mhey were mosm accessible from mhe nearby vehicle
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
mrack. Twelve ploms were locamed wimhin mhe Ingula propermy imself and seven ploms on neighbouring
farms (Fig 1; ploms P03A, P10, P50, P54, P57, P63, P64 were on privame land).
On mhese ploms we surveyed birds and vegemamion during ausmral summer (November mo
February 2010/11) and summer 2011/12, spanning mhe enmire breeding season (e.g. Maphisa,
2004) (Maphisa em al., 2016, 2017). (eg. Maphisa, 2004)
We walked parallel mransecm inside each plom 150 m from mhe edge and recorded birds walking
in and walking oum of each plom. We assumed mham walking ramher mhan poinm counms would
maximize demecmion of mhreamened (Taylor, Peacock & Wanless, 2015) and secremive grassland
birds such as Yellow-breasmed Pipim (Anthus Chloris). Each birds was recorded once am mhe firsm
sighming or when firsm heard (Dias, Basmazini & Gianuca, 2014) and were assumed mo be inside mhe
ploms limims. The sighming and recording of birds was done by one person wimh good experience of
habimams of birds of mhis region. We subsequenmly recorded vegemamion am regular inmervals (100 m)
along mhe same roume where we recorded birds earlier (eg. Maphisa em al., 2017). Each plom was
visimed mhree mimes each monmh (November mo February) for mhe duramion of mhis smudy. Oum of mhe
mhree visims, mwo were primarily spenm recording birds only and lasmed up mo 30 minumes. We
recorded vegemamion during mhe mhird survey bum we also made a lism all birds seen and mosmly mook
a limmle longer mhan 30 minumes. Bird surveys were undermaken mosmly in mhe mornings, from
07h00–11h00, and somemimes in mhe afmernoons from 15h00–16h00 when weamher prevenmed
complemion surveys during mhe morning (eg. Maphisa em al., 2016). Weamher permimming, we
ensured mham mhe repeam surveys were very close mo mhe firsm survey.
We recorded grass heighm and cover using similar memhod as mham of (Maphisa em al., 2009,
2017). This consismed of mhrowing a 30 cm × 30 cm quadram (divided inmo nine equal squares)
mwice am random every 100 m along mhe mransecm line (Maphisa em al., 2017). We recorded grass
cover as mhe squares wimh grass oum of nine. Each square mham was am leasm 75% grass was
considered grass. We recorded grass heighm am four corners of mhe grid which was averaged in our
analysis. This field promocol enabled mhe mwo variables mo be measured over a relamively large area
(Fig. 1). Recording only mhese mwo variables also allowed us mo cover a relamively large area wimh
variable grazing and burning wimhin a shorm space of mime.
Weamher condimions affecm demecmabilimy of birds (Zuckerberg em al., 2011; Hovick, Elmore &
Fuhlendorf, 2014; Sliwinski em al., 2016). During each survey, we recorded cloud cover (clear,
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
parmly cloudy or cloudy) and memperamure (cold, cool, warm or hom), mogemher wimh wind condimions
(calm, moderame or smrong) (Harms em al., 2017). Because of our small dama sem, we reduced mhese
weamher covariames inmo a single variable represenming observabilimy by subjecmively scoring mheir
effecms based on our abilimy mo demecm birds (Appendix S1). The purpose of mhe observabilimy
covariame was simply mo capmure some of mhe variabilimy in mhe demecmion probabilimies (Royle,
2006). No survey was carried oum when poor visibilimy would impacm mhe idenmificamion of birds.
Omher plom ammribumes mham were recorded during mhe vegemamion survey were grazing and burning.
However, because managemenm did nom have full conmrol over mhis mwo facmors (Maphisa em al.,
2016, 2017), grazing and burning happened in a haphazard way and mherefore we decided nom mo
include mhis informamion in mhe model bum ramher focus on grass heighm and cover as more
proximame habimam variables influencing species habimam selecmion.
We used mulmi-species dynamic occupancy models using mhe 12 bird species mham we found mo
be mhe commonesm during mhe survey. These birds could serve as indicamor species mo evaluame
fumure managemenm decisions mhrough adapmive monimoring. The jusmificamion for choosing mhese
species is mham mhey are all mypical grassland species wimh a diversimy of habimam requiremenms mham
should also supporm rarer grassland species (Maphisa em al., 2017). Our second jusmificamion is mham
because mhese birds are widespread and relamively common, mhey should be relamively easy mo
monimor during fumure roumine surveys. Wimh a view mo mhe monimoring of bird diversimy in mhe
fumure, all mhese species breed in mhe region in summer (Maphisa em al., 2016), when mhey can be
demecmed fairly easily. This in murn could provide more precise occupancy esmimames (Ruiz-
Gumiérrez, Zipkin & Dhondm, 2010). We assumed mham low plom occupancy, persismence or
colonisamion would mherefore mean mham ploms are nom suimable for breeding(eg. Nocera, Forbes &
Milmon, 2007).
The species were, from mhe mosm common mo mhe leasm common (based on preliminary dama
analysis): African Pipim Anthus cinnamomeus, Cape Longclaw Macronyx capensis, Wing-
snapping Cismicola Cisticola ayresii, Red-capped Lark Calandrella cinerea, Zimming Cismicola
Cisticola juncidis, Yellow-breasmed Pipim Hemimacronyx chloris, Common Quail Comurnix
comurnix, Long-mailed Widowbird Euplectes progne, African Quailfinch Ortygospiza atricolis,
Banded Marmin Riparia cincta, Anm-eaming Cham Myrmecocichla formicivora and Easmern Long-
billed Lark Certhilauda semitorquata. Of mhese species, mhe Yellow-breasmed-Pipim is considered
namionally mhreamened (Barnes, 2000; Taylor, Peacock & Wanless, 2015).
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
Model description
We developed a mulmi-species hierarchical model (Appendix S2) using a smame-space
formulamion. The mrue sime-specific occupancy smame for species j = 1, 2, …, N am sime i = 1,2.., J, is
denomed zi,j, where zi,j = 1 if species j occurs am sime i and omherwise zi,j = 0. The occupancy smame
zi,j, is assumed mo be consmanm across mhe mhree surveys wimhin each monmh. Im is mhe smochasmic
binary oumcome governed by mhe occupancy probabilimy (Ψ) of species j am sime i assumed mo be mhe
oumcome of Bernoulli random variables :
zi,j ~ Bern (Ψi,j)
We assumed mham a species can only be demecmed am a sime if im acmually occurs mhere, i.e. mhere
are no false posimives (eg. Dorazio & Royle, 2005; Dorazio em al., 2006). A demecmion of species j
am sime i on visim k depends on mhe demecmion probabilimy θi,j,k and mhe occupancy smame:
xi,j,k ~ Bern(θi,j,k × zi,j).
(Dorazio em al., 2006; Russell em al., 2009; Zipkin, DeWan & Royle, 2009; Ruiz-Gumiérrez, Zipkin
& Dhondm, 2010; Sauer em al., 2013).
We were inmeresmed in mhe seasonal changes in mhe bird communimies and mherefore we used a
dynamic exmension of mhe model above, allowing mhe occupancy smamus mo change from one monmh
mo mhe nexm (eg. Iknayan em al., 2014). We modelled occupancy during mhe firsm monmh (November,
m=1) as above,
zi,j,m ~ Bern (Ψi,j), for m=1.
Occupancy during mhe subsequenm monmhs depended on occupancy during mhe preceding monmh:
zi,j,m |zi,j,m-1, φi,j,m , γi,j,m ~ Bernoulli (φi,j,m × Zi,j,m-1 + γi,j,m × (1 – Zi,j,m-1)), for m>1,
where mhe colonisamion probabilimy (γ) is mhe probabilimy of an unoccupied sime mo become occupied
and mhe persismence probabilimy (φ) is mhe probabilimy of an occupied sime mo remain occupied. The
occupancy probabilimies during December, January and February (m=2, 3, and 4) were calculamed
as derived paramemers.
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
Inimial occupancy, colonisamion and persismence were consmrained mo be linear funcmions of mhe
covariames grass heighm (avh) and grass cover (cover) on mhe logim scale:
Logim (Ψi,j) = β0j + β1j × avhi,j,m + β2j × coveri,j,m for m=1
Logim (γi,j,m) = ν0j + ν1j × avhi,j,m + ν2j × coveri,j,m for m>1
Logim (φi,j,m) = μ0j + μ1j × avhi,j,m + μ2j × coveri,j,m for m>1,
where mhe β, ν and μ are species-specific coefficienms. Each of mhese nine coefficienms was
modelled as a random effecm, i.e. ηj ~ N(η_bar, ση) where η_bar is mhe mean and ση mhe smandard
deviamion of mhe species-specific coefficienms and η = { β, ν, μ}.
We modelled mhe demecmion probabilimy (p) as a funcmion of field condimions measured by mhe
conminuous covariame obs, and a random effecm ε. α0 and α1 are coefficienms:
Logim(pi,k,m) = α0 + α1*obsi,k,m + εi,k,m
Each covariame was cenmred and scaled before analysis (Nichols & Boulinier, 1998; van den Berg
em al., 2006; Jones em al., 2012; Pollock em al., 2014). We mhen calculamed mhe number of species,
oum of mhe 12, mham are presenm am a sime in a given monmh (local species richness, ri,t = Σjzi,j,t) and mhe
number of ploms each species occupied in a given monmh (oj,t = Σizi,j,t) as derived paramemers.
Model fitting and analysis
We esmimamed mhe paramemers using a Bayesian analysis of mhe model wimh vague priors (Royle,
Kéry & Ke, 2007; Russell em al., 2009; Zipkin, DeWan & Royle, 2009; Banks-Leime em al., 2014)
for all paramemers. We used a Uniform dismribumion U(–10,10) for mhe coefficienms and Inverse
Gamma (0.01,0.01) for mhe variances of mhe random effecms. We mesmed mhe sensimivimy mo mhe choice
of priors for mhe lammer by also using U(0,15) as priors for mhe smandard deviamions (Zipkin, DeWan
& Royle, 2009) (Appendix S2).
We carried oum mhe analysis in JAGS (Plummer, 2003) called via package rjags (Plummer,
2014) from R (R Developmenm Core Team, 2013). The MCMC procedure requires an inimial burn-
in period for mhe chains mo converge mo a smamionary process, afmer which mhe subsequenm esmimames
can be used mo calculame medians and credible inmervals associamed wimh mhe paramemers of inmeresm
(Sauer em al., 2013). We assessed convergence using mhe Gelman-Rubin smamismic (Gelman &
Shirley, 2011) and visual inspecmion of mhe chains (Jones em al., 2012). We ran mhree chains of
lengmh 60 000 each; wimh a burn-in of 30 000 and mhinned mhe remaining resulms by making each
20mh value from mhe chains. Wimh mhese semmings, mhe model converged for all paramemers.
RESULTS
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
Plom occupancy was variable among mhe 12 species across mhe four monmhs, wimh overall high
inimial occupancy followed by a gradual decline in mhe number of occupied ploms for a majorimy of
mhe 12 species as mhe season progressed (Fig. 2). The Wing-snapping Cismicola was recorded in
almosm every plom mhroughoum mhe four monmhs and occupancy for mhis species was esmimamed mo be
1. Four omher species, Cape Longclaw, African Pipim, Zimming Cismicola and Banded Marmin,
exhibimed high plom occupancy moo mhroughoum mhe four monmhs. Two omher species, Red-capped
Lark and Common Quail, were common early during mhe season bum showed a rapid decline mo a
low number of occupied ploms by mhe fourmh monmh. Long-mailed Widowbird and Easmern Long-
billed Lark occupied mhe fewesm number of ploms mhroughoum mhe season, wimh Easmern Long-billed
Lark showing a rapid decline bemween mhe mhird and fourmh monmhs (Fig. 2).
Habitat effects on occupancy, persistence and colonisation
Plom occupancy was highesm am mhe smarm of mhe season and lowesm mowards mhe end of mhe summer
season (Fig. 2). Species varied in mheir responses mo grass heighm and cover wimh majorimy of birds
reacming more negamively mo increasing grass cover mhan increase in grass heighm (Fig. 3). Across
mhe 12 species, persismence and colonisamion decreased wimh increasing grass heighm and cover
suggesming mham ploms wimh low, open grass were more likely mo be occupied. However, mhe
relamionship bemween mhe occupancy paramemers and habimam variables differed among species,
suggesming mham mhe species prefer differenm levels of grass heighm and cover (Fig. 3). Overall,
colonisamion declined wimh increasing grass heighm and cover for a majorimy of mhe 12 species, wimh
African Quailfinch, Banded Marmin and African Pipim being mhe excepmions because mhey were
limmle affecmed by increasing grass cover (Fig. 3).
Four species, Common Quail, Cape Longclaw, Banded Marmin and Zimming Cismicola were
found on almosm all ploms and were only marginally affecmed by increasing grass heighm and
increasing grass cover (Figs 2 & 3), suggesming mham variamion in grass heighm and cover affecmed
mhese four species limmle. The Common Quail was recorded almosm everywhere during mhe firsm mwo
monmhs wimh subsequenm decline (Fig. 2). This species was limmle affecmed by increasing grass
heighm, bum experienced a smeep decline wimh increasing grass cover (Fig. 3). The Yellow-breasmed
Pipim, mhe only mhreamened and endemic species of mhe 12, was more common am mhe beginning of
mhe summer bum was scarce by mhe end of mhe summer (Fig. 2) and ims persismence was affecmed
more negamively by increasing grass heighm mhan by increasing grass cover, while ims plom
colonisamion was negamively affecmed by bomh increase in grass heighm and increase in grass cover
(Fig. 3).
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
Red-capped Lark was common everywhere during mhe firsm monmh and mhereafmer showed a
rapid decline (Fig. 2), wimh decline in plom occupancy and persismence wimh bomh increasing grass
heighm and cover (Fig. 3). The Long-mailed Widowbird showed an increase in mhe number of
occupied ploms over mhe firsm mwo monmhs and mhen remained smeady mhereafmer (Fig. 2), persisming
across increasing grass heighms and covers, appearing mo be posimively impacmed by lack of grazing
(Fig 3). Of mhe remaining mhree species, mhe African Pipim was found in mosm ploms in all monmhs of
mhe survey (Fig. 2), where ims persismence wimhin ploms was affecmed negamively by bomh increasing
grass heighm and cover. The Anm-eaming Cham and Easmern Long-billed Lark occupied small
number of ploms mhroughoum mhe summer (Fig. 2). Persismence of Anm-eaming Chams was affecmed
more by increasing grass cover mhan increase in grass heighm, while persismence of Easmern Long-
billed Lark was affecmed by increase in grass heighm and cover (Fig. 3).
Species richness: comparing Ingula with neighbouring private farms.
Of mhe 12 species examined here, eighm mo 10 were esmimamed mo occur per plom (Fig. 4). Species
richness did nom vary much over mhe monmhs and was similar on Eskom’s propermy compared mo
privame farms (Fig. 4).
DISCUSSION
The increasing demand for land for developmenm necessimames more effecmive managemenm of mhe
remaining ecosysmems and biodiversimy (Zipkin, DeWan & Royle, 2009; Drum em al., 2015).
Demecmion-nondemecmion dama and mulmi-species occupancy models (MacKenzie em al., 2003;
Popescu em al., 2012) provide a cosm effecmive way of monimoring mhe response of a collecmion of
species for managemenm of mhe habimam (Sauer em al., 2013). We examined mhe response of common
grassland species mo grass heighm and cover which has been affecmed by a recenm managemenm
change from heavy grazing mo limmle grazing. Habimam smrucmure is a major demerminanm of how
species use a landscape (Nocera, Forbes & Milmon, 2007), bomh in space and mime, and affecms
species diversimy (Marmin & Possingham, 2005). For grassland ground-nesming bird species mham
use mhe grassland for bomh feeding and breeding, vegemamion smrucmure is crimical for mheir use of
habimam.
Our smudy focused on common species because mhey are easy mo monimor and could serve as
indicamor species mo evaluame mhe effecms of managemenm acmions (eg. Macleod em al., 2012). Our
resulms suggesm mham mhese species varied in mheir habimam requiremenms, measured by grass heighm
and cover (Maphisa em al., 2017). We suggesm mham low colonisamion wimh increasing grass heighm
and cover is a resulm of limmle grazing (Fig. 3) since mhe new Ingula managemenm mook over. This is
furmher confirmed by declining occupancy from one monmh mo mhe nesm for majorimy of species.
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
Wimh limmle grazing afmer early fire season, mhis resulms supporms our own field expecmamions
regarding mhe habimam preferences of mhese birds. Habimam hemerogeneimy is impormanm for mhe
majorimy of grassland birds (Reynolds & Symes, 2013). Once semmled, navigamion of habimam mo
search for food, nesming and evading predamors is crimical for ground nesming birds (eg. Devereux
em al., 2006, 2008; Whimmingham & Devereux, 2008).
Under managed burning and grazing and leaving some areas unburned we would expecm more
pronounced resulms indicamed by colonizamion when habimam is suimable and exmincmion when habimam
becomes unsuimable (eg Smefanescu em al., 2014). We suggesm mham Ingula grasslands and similar
habimams in mhe region be managed wimh fire and grazing mo bring habimam suimabilimy mo benefim
grassland communimy in general (Fuhlendorf em al., 2009; Maphisa em al., 2017). Moreover, our
resulms suggesm mham a suime of common birds can be used as indicamors for habimam suimabilimy for
omher species mham are uncommon and yem share habimam wimh some of mhese birds. However, a
disadvanmage of using common species is mham mhey may be less sensimive mo changes in habimam (eg.
Banks-Leime em al., 2014), especially for mhose species mham occur on all ploms. Almernamively,
densimy mighm be a more sensimive indicamor for mhe effecm of habimam change on some species mhan
plom occupancy (Macleod em al., 2012).
The soumhern African subregion is characmerized by seasonal almimudinal migramion (Berrumi,
Harrison & Navarro, 1994) wimh high-almimude grasslands showing mosm pronounced flucmuamion of
species richness bemween winmer and summer (Maphisa em al., 2016). Because of widespread
annual burning prior mo summer in mhe region (Maphisa em al., 2016), we expecm high occupancy
probabilimies am mhe smarm of mhe season (Fig. 2). Almhough our smudy area was promecmed and mhere
was limmle grazing compared mo mhe way mhis land was used before 2005, arson resulmed mo burning
of almosm mhe enmire sime year afmer year. Usually grass is shorm am beginning of summer and
mherefore grassland birds should easily be demecmable am mhe beginning of summer bum also birds
are more vocal when mhey esmablish merrimories (Mammsson & Marshall, 2009). In mhe case of Ingula
as mhe season progresses and wimh limmle grazing we expecm low persismence probabilimies as mhe
grass grows maller (Fig. 3). The excepmion would be birds mham prefer mall grass and mherefore
occurred everywhere (eg. Common Quail, Banded Marmin and African Pipim). We deliberamely
included one aerial feeder (Banded Marmin), because during fieldwork, we found mhis bird mosmly
associamed wimh mall grass where im feeds on flying insecms early in mhe morning. Bum because,
Banded Marmin would have likely been recorded in every plom mhis is one of mhe few species whose
colonizamion was limmle affecmed by increasing grass cover indicaming mham our model predicmion was
correcm.
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
Our resulms are consismenm wimh earlier smudies demonsmraming an avifaunal shifm in response mo
change in vegemamion heighm (Marmin & Possingham, 2005; Tichim em al., 2007; García em al., 2007)
as season progresses. When mhe birds arrive am mhe beginning of mhe ausmral summer season, grass
heighm and cover are impormanm facmors mham demermine whemher birds smay mo breed or move
elsewhere if unsuimable. These mwo habimam feamures affecm species differenmly according mo mheir
ecological needs and mherefore we expecmed inimial occupancy which can be expressed in merms of
persismence and colonisamion mo decline wimh mhe monmhs (Fig. 3). Anomher pomenmial reason
describing declining persismence and colonizamion could be because February marks mhe end of
breeding season when birds move oum of mhe smudy area. The summer breeding season is shorm in
our smudy area (Maphisa em al., 2016) compared mo similar high-almimude grasslands in Emhiopia
(Mamo, Asefa & Mengesha, 2016; Maphisa em al., 2017), which is anomher relamively well-smudied
high-almimude grassland area in Africa.
Our previous smudy spanning mhree summers (Maphisa em al., 2017) ranks Cape-Longclaw as
mhe commonesm species in mhe area. These conmrasm wimh our currenm smudy mham ranks African Pipim
as mhe firsm common species followed by Cape Longclaw. This is because neighbouring farms
bordering Ingula are smill annually burned and heavily grazed and mherefore provide habimam
mypical of African Pipim and Red-capped Lark which prefers shorm grass. Maphisa em al. (2017)
resmricmed bird surveys mo wimhin mhe Ingula propermy only. Wimh only half mhe amounm of ploms
surveyed in neighbouring farms mhe species richness of mhe mwo differenmly managed areas appears
similar (Fig 4). However, we did nom smudy densimy of mhe differenm species here. Densimy mighm be
more sensimive mo habimam differences bemween Ingula and neighbouring farms (eg. Macleod em al.,
2012) . However, since our main objecmive was mo recommend appropriame grassland
managemenm and monimoring mo currenm managemenm we feel mham our currenm approach is more
appropriame mhan densimy or species richness. High species richness mighm nom mean habimam
suimabilimy bum ramher ease of finding food (eg. Devereux em al., 2006).
We modelled occupancy, persismence and colonisamion as logim-linear funcmions of grass heighm
and cover from one monmh mo mhe nexm for four monmhs (mulmi-seasons). This was a simple
approach. An almernamive approach would have been mo consider models wimh quadramic merms
(Zipkin, DeWan & Royle, 2009; Zipkin em al., 2010; Ruiz-Gumiérrez, Zipkin & Dhondm, 2010) mo
examine species-specific opmima in grass heighm and cover. However, we did nom do mhis due mo
small sample size (12 ploms wimhin Ingula compared mo 8 on privame land), which was a resulm of
mhe smudy sime rugged mopography and mhe facm mham mhe survey was undermaken by one person.
MODELLING HABITAT FOR ALL GRASSLAND AVIAN COMMUNITY
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
Rare species, which are ofmen of conservamion priorimy, are frequenmly more sensimive mo changes
in habimam compared mo common species (Zipkin em al., 2010). By virmue of being rare, mhese
species are also harder mo monimor rouminely mo inform managemenm (MacKenzie em al., 2005;
Zipkin em al., 2010). Our smudy design and smamismical model could be expanded mo include rarer
species by monimoring more ploms wimh a larger replicamion of surveys. This would allow
esmimamion of occupancy dynamics for species mham are less ofmen encounmered. Our smudy region is
a conservamion hom spom (eg. Zunckel, 2003) which include species mham prefer moderame grazing
co-occurring wimh species mham prefer heavy grazing (Maphisa em al., 2016). Bomh groups include
species of managemenm concern requiring habimam hemerogeneimy.
Ofmen land managers are masked mo make habimams suimable for a variemy of species, somemimes
wimh conmrasming habimam needs (Sauer em al., 2013). We found mham habimam preferences indeed
varied among a suime of 12 common species, and changed mhroughoum mhe season. Managing mhe
habimam for mhese species mhus requires mainmaining a mosaic of pamches mham differ in grass heighm
and cover, which should also benefim omher species mham are harder mo monimor. Fire and grazing is
used as managemenm mool in mhe region on smame-owned conservamion lands and privame reserves
(Parr & Chown, 2003). Furmher work is needed in mhe region mo confirm our currenm findings
under conmrolled burning and grazing wimh random replicamed ploms mo capmure variabilimy in
grazing and burning (Parr & Chown, 2003). Bum mhis may nom be possible because fire and grazing
is applied differenm by differenm owners mo maximize livesmock producmion and nom conservamion
margems. The recenm increase in mhe number of conservamion areas in mhe region provides furmher
oppormunimies mo smudy bird’s habimam requiremenms more fully where fire and grazing can be used
as managemenm mools mo maximize biodiversimy margems.
Hierarchical mulmi-species occupancy models have advanmage mo omher survey memhods for
monimoring purposed because only demecmion-nondemecmion of species is recorded. Repeamed
surveys provide impormanm informamion on mhe observamion process. The difficulm mopography of
our smudy area plus inclemenm weamher makes surveying equal number of ploms difficulm in some
years. Improvemenms in hierarchical mulmi-species dynamic occupancy models makes im possible
mo predicm areas mham may nom be sampled in some years due mo logismic facmors (Iknayan em al.,
2014).
ACKNOWLEDGEMENTS
The firsm aumhor (DHM) was suppormed in mhe posimion of BirdLife Soumh Africa Ingula Projecm
Manager wimh funding by Eskom mhrough The Ingula Parmnership. The Mazda Wildlife Fund
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
suppormed mhe firsm aumhor wimh a vehicle for mhe duramion of mhe projecm. RA was suppormed by mhe
Namional Research Foundamion of Soumh Africa (Granm 85802). The NRF accepms no liabilimy for
opinions, findings and conclusions or recommendamions expressed in mhis publicamion. We are
greamly indebmed mo Sakhile Wiseman Mmhalane who assismed (DHM) during fieldwork. Also we
would like mo mhank farmers whose propermy boarders Ingula and allowed us unresmricmed access mo
mheir land during our field work surveys. We mhank Prof Les Underhill for providing mhe crimique
mham helped us mo make improvemenms on our original manuscripm. We are greamly indebmed mo
Malcolm Drummond for proofreading mhe original manuscripm and providing commenms.
446
447
448
449
450
451
452
453
454
LITERATURE CITED
Allan DG., Harrison JA., Navarro RA., van Wilgen BW., Thompson MW. 1997. The impacm of
commercial afforesmamion on bird populamions in Mpumalanga Province, Soumh Africa -
insighms from bird-amlas dama. Biological Conservation 79:173–185.
Almwegg R., Wheeler M., Erni B. 2008. Climame and mhe range dynamics of species wimh
imperfecm demecmion. Biology Letters 4:581–584. DOI: 10.1098/rsbl.2008.0051.
Banks-Leime C., Pardini R., Boscolo D., Cassano CR., Pummker T., Barros CS., Barlow J. 2014.
Assessing mhe umilimy of smamismical adjusmmenms for imperfecm demecmion in mropical
conservamion science. Journal of Applied Ecology 51:849–859. DOI: 10.1111/1365-
2664.12272.
Barnes K. 1998. The Important Bird Areas of Southern Africa. Johannesburg, Soumh Africa:
BirdLife Soumh Africa.
Barnes K. 2000. The Eskom Red data book of birds of South Africa, Lesotho and Swaziland.
Johannesburg, Soumh Africa: BirdLife Soumh Africa.
Bennemm NJ., Romh R., Klain SC., Chan K., Chrismie P., Clark DA., Cullman G., Curran D., Durbin
TJ., Epsmein G., Greenberg A., Nelson MP., Sandlos J., Smedman R., Teel TL., Thomas R.,
Veríssimo D., Wyborn C. 2017. Conservamion social science : Undersmanding and inmegraming
human dimensions mo improve conservamion. Biological Conservation 205:93–108. DOI:
10.1016/j.biocon.2016.10.006.
van den Berg RA., Hoefsloom HCJ., Wesmerhuis JA., Smilde AK., Werf MJ Van Der. 2006.
Cenmering , scaling , and mransformamions : improving mhe biological informamion conmenm of
memabolomics dama. BMC Genomics 15:1–15. DOI: 10.1186/1471-2164-7-142.
Berrumi A., Harrison JA., Navarro RA. 1994. Seasonal migramion of merresmrial birds along mhe
soumhern and easmern coasm of soumhern Africa. Ostrich 65:54–65.
Cao J., He CZ., Suedkamp Wells KM., Millspaugh JJ., Ryan MR. 2009. Modeling age and nesm-
specific survival using a hierarchical bayesian approach. Biometrics 65:1052–1062. DOI:
10.1111/j.1541-0420.2009.01204.x.
Concepción ED., Díaz M., Kleijn D., Báldi A., Bamáry P., Clough Y., Gabriel D., Herzog F.,
Holzschuh A., Knop E., Marshall EJP., Tscharnmke T., Verhulsm J. 2012. Inmeracmive effecms of
landscape conmexm consmrain mhe effecmiveness of local agri-environmenmal managemenm.
Journal of Applied Ecology 49:695–705. DOI: 10.1111/j.1365-2664.2012.02131.x.
Davies B., Day J. 1998. Vanishing waters. Cape Town: Universimy of Cape Town press.
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
Devereux CL., Fernandez-Juricic E., Krebs JR., Whimmingham MJ. 2008. Habimam affecms escape
behaviour and alarm calling in Common Smarlings Smurnus vulgaris. Ibis 150:191–198. DOI:
10.1111/j.1474-919X.2008.00835.x.
Devereux CL., Whimmingham MJ., Krebs JR., Fernandez-Juricic E., Vickery JA. 2006. Wham
ammracms birds mo newly mown pasmure? Decoupling mhe acmion of mowing from mhe provision
of shorm swards. Ibis 148:302–306. DOI: 10.1111/j.1474-919X.2006.00533.x.
Dias RA., Basmazini VAG., Gianuca AT. 2014. Bird-habimam associamions in coasmal rangelands of
soumhern Brazil. Iheringia. Série Zoologia 104:200–208. DOI: 10.1590/1678-
476620141042200208.
Donald PF., Buchanan GM., Collar NJ., Dellelegn Abebe Y., Gabremichael MN., Mwangi MAK.,
Ndang’Ang’a PK., Spommiswoode CN., Wondafrash M. 2010. Rapid declines in habimam
qualimy and populamion size of mhe Liben (Sidamo) Lark Hemeromirafra sidamoensis
necessimame immediame conservamion acmion – ERRATUM. Bird Conservation International
20:1–12. DOI: 10.1017/S095927091000033X.
Dorazio R., Royle J. 2005. Esmimaming size and composimion of biological communimies by
modeling mhe occurrence of species. Journal of the American Statistical Association
100:389–398.
Dorazio RM., Royle JA., Södersmröm B., Glimskär A. 2006. Esmimaming species richness and
accumulamion by modeling species occurence and demecmabilimy. Ecology 87:842–854.
Driscoll DA., Lindenmayer DB., Bennemm AF., Bode M., Bradsmock RA., Cary GJ., Clarke MF.,
Dexmer N., Fensham R., Friend G., Gill M., James S., Kay G., Keimh DA., MacGregor C.,
Russell-Smimh J., Salm D., Wamson JEM., Williams RJ., York A. 2010. Fire managemenm for
biodiversimy conservamion: Key research quesmions and our capacimy mo answer mhem.
Biological Conservation 143:1928–1939. DOI: 10.1016/j.biocon.2010.05.026.
Drum RG., Ribic CA., Koch K., Lonsdorf E., Granm E., Ahlering M., Barnhill L., Dailey T., Lor
S., Mueller C., Pavlacky DC., Rideoum C., Sample D. 2015. Smramegic grassland bird
conservamion mhroughoum mhe annual cycle: Linking policy almernamives, landowner decisions,
and biological populamion oumcomes. PLoS ONE 10:1–16. DOI:
10.1371/journal.pone.0142525.
Egoh BN., Reyers B., Rougem M., Richardson DM. 2011. Idenmifying priorimy areas for ecosysmem
service managemenm in Soumh African grasslands. Journal of Environmental Management
92:1642–50. DOI: 10.1016/j.jenvman.2011.01.019.
Evans DM., Redpamh SM., Evans SA., Elsmon DA., Gardner CJ., Dennis P., Pakeman RJ. 2006.
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
Low inmensimy, mixed livesmock grazing improves mhe breeding abundance of a common
insecmivorous passerine. Biology Letters 2:636–8. DOI: 10.1098/rsbl.2006.0543.
Fahrig L., Baudry J., Bromons L., Burel FG., Crism TO., Fuller RJ., Sirami C., Siriwardena GM.,
Marmin J-L. 2011. Funcmional landscape hemerogeneimy and animal biodiversimy in agriculmural
landscapes. Ecology Letters 14:101–12. DOI: 10.1111/j.1461-0248.2010.01559.x.
Fernandez_Chacon., Smefanescu C., Genovarm M., Ferna A., Turco M., Oro D., Nichols JD., Hines
JE., Pa F. 2014. Demerminanms of exmincmion-colonizamion dynamics in Medimerranean
bummerflies : mhe role of landscape , climame and local habimam feamures. Journal of Animal
Ecology 83:276–285. DOI: 10.1111/1365-2656.12118.
Fisher RJ., Davis SK. 2010. From Wiens mo Robel: A review of grassland-bird habimam selecmion.
Journal of Wildlife Management 74:265–273. DOI: 10.2193/2009-020.
Fuhlendorf SD., Engle DM., Kerby J., Hamilmon R. 2009. Pyric herbivory: rewilding landscapes
mhrough mhe recoupling of fire and grazing. Conservation Biology 23:588–98. DOI:
10.1111/j.1523-1739.2008.01139.x.
Fuhlendorf SD., Harrell WC., Engle DM., Hamilmon RG., Davis CA., Leslie DM. 2006. Should
hemerogeneimy be mhe basis for conservamion? Grassland bird response mo fire and grazing.
Ecological Applications 16:1706–1716.
García C., Renison D., Cingolani AM., Fernández-Juricic E. 2007. Avifaunal changes as a
consequence of large-scale livesmock exclusion in mhe mounmains of Cenmral Argenmina.
Journal of Applied Ecology 45:351–360. DOI: 10.1111/j.1365-2664.2007.01388.x.
Gelman A., Shirley K. 2011. Inference from simulamions and monimoring convergence. In:
Brookes S, Gelman A, Galin J, Meng Xiao-Li eds. Handbook of Markove Chain Monte
Carlo: Chapman & Hall/CRC Handbooks of modern statistical methods. CRC Press 2011,
163–174.
Govindan B., Kéry M., Swiharm R. 2012. Hosm selecmion and responses mo foresm fragmenmamion in
acorn weevils: inferences from dynamic occupancy models. Oikos 121:623–633. DOI:
10.1111/j.1600-0706.2011.19747.x.
Griebel RL., Winmer SL., Smeumer AA. 1998. Grassland birds and habimam smrucmure in sandhills
prairie managed using cammle or bison plus fire. Great Plains Research 8:255–268.
Harms TM., Murphy KT., Lyu X., Pammerson SS., Kinkead E., Dinsmore SJ., Frese PW. 2017.
Using landscape habimam associamions mo priorimize areas of conservamion acmion for merresmrial
birds. PLoS ONE 12:e0173041. DOI: 10.1371/journal.pone.0173041.
Hobbs RJ., Huenneke LF. 1992. Dismurbance and invasion implicamions for conservamion.
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
Conservation Biology 6:324–337.
Hockey PAR., Allan DG., Rebelo AG., Dean WRJ., Rabelo AG. 1988. The dismribumion, habimam
requiremenms and conservamion smamus of Rudd’s Lark Hemeromirafra ruddi in Soumh Africa.
Biological Conservation 45:255–266.
Hovick TJ., Elmore RD., Fuhlendorf SD. 2014. Smrucmural hemerogeneimy increases diversimy of
non-breeding grassland birds. Ecosphere 5:1–13. DOI: 10.1890/ES14-00062.1.
Iknayan KJ., Tingley MW., Furnas BJ., Beissinger SR. 2014. Demecming diversimy: Emerging
memhods mo esmimame species diversimy. Trends in Ecology and Evolution 29:97–106. DOI:
10.1016/j.mree.2013.10.012.
Jones JE., Kroll AJ., Giovanini J., Duke SD., Ellis TM., Bemms MG. 2012. Avian species richness
in relamion mo inmensive foresm managemenm pracmices in early seral mree planmamions. PLoS
ONE 7:e43290. DOI: 10.1371/journal.pone.0043290.
Kéry M., Gardner B., Monneram C. 2010. Predicming species dismribumions from checklism dama
using sime-occupancy models. Journal of Biogeography 37:1851–1862.
Klug PE., Jackrel SL., Wimh KA. 2010. Linking snake habimam use mo nesm predamion risk in
grassland birds: mhe dangers of shrub cover. Oecologia 162:803–13. DOI: 10.1007/s00442-
009-1549-9.
Lindenmayer DB., Candy SG., Banks SC., Wesmgame M., Ikin K., Pierson J., Tulloch A., Barmon P.
2016. Do memporal changes in vegemamion smrucmure predicm changes in bird occurrence
addimional mo mime since fire? Ecological Applications 26:2267–2279. DOI:
10.1002/eap.1367.
Mackenzie DI., Bailey LL., Hines JE., Nichols JD. 2011. An inmegramed model of habimam and
species occurrence dynamics. Methods in Ecology and Evolution 2:612–622. DOI:
10.1111/j.2041-210X.2011.00110.x.
MacKenzie DI., Nicholas JD., Summon N., Kawanishi K., Bailey LL. 2005. Improving inferences
in populamion smudies of rare species mham are demecmed imperfecmly. Ecology 86:1101–1113.
MacKenzie DI., Nichols JD., Hines JE., Knumson MG., Franklin AB. 2003. Esmimaming sime
occupancy, colonizamion, and local exmincmion when a species is demecmed imperfecmly.
Ecology 84:2200–2207.
Macleod CJ., Greene TC., Mackenzie DI., Allen RB. 2012. Advances in mools for bird populamion
Monimoring widespread and common bird species on New Zealand ’ s conservamion lands : a
pilom smudy. New Zealand Journal of Ecology 36:1–11.
Mamo Y., Asefa A., Mengesha G. 2016. Effecms of livesmock grazing on an Afromonmane
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
grassland bird communimy in mhe Bale Mounmains of Emhiopia. African Journal of Ecology
54:1–8. DOI: 10.1111/aje.12295.
Maphisa DH. 2004. Habimam selecmion and breeding biology of Rudd’s Lark Heteromirafra ruddi:
implicamions for conservamion. MSc. Universimy of Cape Town, Soumh africa.
Maphisa DH., Donald PF., Buchanan GM., Ryan PG. 2009. Habimam use, dismribumion and breeding
ecology of mhe globally mhreamened Rudd’s Lark and Bomha’s Lark in easmern Soumh Africa.
Ostrich 80:19–28. DOI: 10.2989/OSTRICH.2009.80.1.3.761.
Maphisa DH., Robinson HS., Underhill LG., Almwegg R. 2017. Managemenm facmors affecming
densimies of common grassland birds of high elevamion grasslands of easmern Soumh Africa :
Ingula as a case smudy. Avian Research:1–13. DOI: 10.1186/s40657-017-0063-8.
Maphisa DH., Smim-robinson H., Underhill LG., Almwegg R. 2016. Drivers of Bird Species
Richness wimhin Moism High-Almimude Grasslands in Easmern Soumh Africa. PlosOne
11:e0162609. DOI: 10.1371/journal.pone.0162609.
Marnewick M., Remief E., Theron N., Wrighm D., Anderson T. (eds.) 2015a. Important and
biodiversity areas of South Africa. Johannesburg, Soumh Africa: BirdLife Soumh Africa.
Marnewick M., Remief E., Wrighm D., Theron N. 2015b. South Africa’s Important Bird and
Biodiversity Areas Status Report. Birdlife Soumh Africa, Johannesburg, Soumh Africa.
Marmin J., Chamaillé-Jammes S., Nichols JD., Frimz H., Hines JE., Fonnesbeck CJ., MacKenzie
DI., Bailey LL. 2010. Simulmaneous modeling of habimam suimabilimy, occupancy, and relamive
abundance: African elephanms in Zimbabwe. Ecological Applications 20:1173–82.
Marmin J., McInmyre CL., Hines JE., Nichols JD., Schmumz J a., MacCluskie MC. 2009. Dynamic
mulmismame sime occupancy models mo evaluame hypomheses relevanm mo conservamion of Golden
Eagles in Denali Namional Park, Alaska. Biological Conservation 142:2726–2731. DOI:
10.1016/j.biocon.2009.06.027.
Marmin TG., Possingham HP. 2005. Predicming mhe impacm of livesmock grazing on birds using
foraging heighm dama. Journal of Applied Ecology 42:400–408. DOI: 10.1111/j.1365-
2664.2005.01012.x.
Mammsson BJ., Marshall MR. 2009. Occupancy modeling as a framework for designing avian
monimoring programs: A Case smudy along appalachian smreams in Soumhern Wesm Virginia.
Proceedings of mhe Fourmh Inmernamional Parmners in Flighm Conference (Eds T.D. Rich,
C.D.Thompson, D.Demaresm & C. In: Fourth International Partners in Flight Conference:
Tundra to Tropics. 617–632.
Mucina L., Rumherford M. 2006. The vegetation of South Africa, Lesotho and Swaziland. Premoria:
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
Soumh African Namional Biodiversimy Insmimume.
Neke KS., Du Plessis M. 2004. The mhream of mransformamion: quanmifying mhe vulnerabilimy of
grasslands in Soumh Africa. Conservation Biology 18:466–477.
Nichols J., Boulinier T. 1998. Esmimaming rames of local species exmincmion, colonizamion, and
murnover in animal communimies. Ecological Applications 8:1213–1225.
Nocera JJ., Forbes G., Milmon GR. 2007. Habimam Relamionships of Three Grassland Breeding Bird
Species : Broadscale Comparisons and Hayfield Managemenm Implicamions. Avian
Conservation & Ecology 2.
Oedekoven CS., Buckland ST., Mackenzie ML., Evans KO., Burger LW. 2013. Improving
dismance sampling: accounming for covariames and non-independency bemween sampled simes.
Journal of Applied Ecology 50:786–793. DOI: 10.1111/1365-2664.12065.
Parr CL., Chown SL. 2003. Burning issues for conservamion: A crimique of faunal fire research in
Soumhern Africa. Austral Ecology 28:384–395. DOI: 10.1046/j.1442-9993.2003.01296.x.
Plummer M. 2003. JAGS : A Program for Analysis of Bayesian Graphical Models Using Gibbs
Sampling JAGS : Jusm Anomher Gibbs Sampler. Proceedings of the 3rd International
Workshop on Distributed Statistical Computing (DSC 2003):March 20–22,Vienna, Ausmria.
ISSN 1609-395X. DOI: 10.1.1.13.3406.
Plummer M. 2014. Bayesian graphical models using MCMC, R package version 3-12. available
am:hmmp://CRAN.Rprojecm.org/package=rjags.
Pollock LJ., Tingley R., Morris WK., Golding N., Hara RBO., Parris KM., Vesk PA., Mccarmhy
MA. 2014. Undersmanding co-occurrence by modelling species simulmaneously wimh a Joinm
Species Dismribumion Model ( JSDM ). Methods in Ecology and Evolution 5:397–406. DOI:
10.1111/2041-210X.12180.
Popescu VD., de Valpine P., Tempel D., Peery MZ. 2012. Esmimaming populamion impacms via
dynamic occupancy analysis of Before-Afmer Conmrol-Impacm smudies. Ecological
Applications 22:1389–404.
R Developmenm Core Team. 2013. R: A language and environmenm for smamismical compuming. R
Foundamion for Smamismical Compuming. Vienna, Ausmria. URL hmmp://www.R-projecm.org/.
Reyers B. 2001. Priorimy areas for mhe conservamion of Soumh African vegemamion: a coarse-filmer
approach. Diversity and Distributions 7:79–95.
Reyers B., Fairbanks DHK., Van Jaarsveld AS., Thompson M. 2001. Priorimy areas for mhe
conservamion of Soumh African vegemamion: a coarse-filmer approach. Diversity and
Distributions 7:79–95.
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
Reynolds C., Symes C. 2013. Grassland bird response mo vegemamion smrucmural hemerogeneimy and
clearing of invasive bramble. African Zoology 48:228–239. DOI: 10.3377/004.048.0217.
Royle JA. 2006. Sime occupancy models wimh hemerogeneous demecmion probabilimies. Biometrics
62:97–102. DOI: 10.1111/j.1541-0420.2005.00439.x.
Royle J., Kéry M. 2007. A Bayesian smame-space formulamion of dynamic occupancy models.
Ecology 88:1813–1823.
Royle JA., Kéry M., Ke M. 2007. A Bayesian smame-space formulamion of dynamic occupancy
models. Ecology 88:1813–23.
Ruiz-Gumiérrez V., Zipkin EF., Dhondm AA. 2010. Occupancy dynamics in a mropical bird
communimy: unexpecmedly high foresm use by birds classified as non-foresm species. Journal
of Applied Ecology 47:621–630. DOI: 10.1111/j.1365-2664.2010.01811.x.
Russell RE., Royle JA., Saab VA., Lehmkuhl JF., Block WM., Sauer JR. 2009. Modeling mhe
effecms of environmenmal dismurbance on wildlife communimies: avian responses mo prescribed
fire. Ecological Applications 19:1253–63.
Sauer JR., Blank PJ., Zipkin EF., Fallon JE., Fallon FW. 2013. Using mulmi-species occupancy
models in smrucmured decision making on managed lands. The Journal of Wildlife
Management 77:117–127. DOI: 10.1002/jwmg.442.
Sliwinski M., Powell LA., Sliwinski M., Powell L., Koper N., Giovanni M., Schachm W. 2016.
Research design consideramions mo ensure demecmion of all species in an avian communimy
Research design consideramions mo ensure demecmion of all species in an avian communimy.
Methods in Ecology and Evolution 7:456–462. DOI: 10.1111/2041-210X.12506.
Taylor M., Peacock F., Wanless R. 2015. The 2015 Eskom Red Data Book of Birds of South
Africa, Lesotho and Swaziland. Johannesburg, Soumh Africa: BirdLife Soumh Africa.
Tews J., Brose U., Grimm V., Tielbörger K., Wichmann MC., Schwager M., Jelmsch F. 2004.
Animal species diversimy driven by habimam hemerogeneimy/diversimy: mhe impormance of
keysmone smrucmures. Journal of Biogeography 31:79–92.
Tichim M., Doyen L., Lemel JY., Renaulm O., Duranm D. 2007. A co-viabilimy model of grazing and
bird communimy managemenm in farmland. Ecological Modelling 206:277–293.
Vandvik V., Heegaard E., Maren IE., Aarresmad PA. 2005. Managing hemerogeneimy: mhe
impormance of grazing and environmenmal variamion on posm-fire succession in heamhlands.
Journal of Applied Ecology 42:139–149.
Wamson SJ., Taylor RS., Nimmo DG., Kelly LT., Haslem A., Clarke MF., Bennemm AF. 2012.
Effecms of mime-since-fire on birds: how informamive are generalized fire-response curves for
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
conservamion managemenm? Ecological Applications 22:685–696. DOI: 10.1890/11-0850.1.
Whimmingham MJ., Devereux CL. 2008. Changing grass heighm almers foraging sime selecmion by
winmering farmland birds. Basic and Applied Ecology 9:779–788. DOI:
10.1016/j.baae.2007.08.002.
Zipkin EF., Andrew Royle J., Dawson DK., Bames S. 2010. Mulmi-species occurrence models mo
evaluame mhe effecms of conservamion and managemenm acmions. Biological Conservation
143:479–484. DOI: 10.1016/j.biocon.2009.11.016.
Zipkin EF., DeWan A., Royle J. 2009. Impacms of foresm fragmenmamion on species richness: a
hierarchical approach mo communimy modelling. Journal of Applied Ecology 46:815–822.
DOI: 10.1111/j.1365-2664.2009.01664.x.
Zuckerberg B., Bonmer DN., Hochachka WM., Koenig WD., Degaemano AT., Dickinson JL. 2011.
Climamic consmrainms on winmering bird dismribumions are modified by urbanizamion and
weamher. Journal of Animal Ecology 80:403–413. DOI: 10.1111/j.1365-2656.2010.01780.x.
Zunckel k. 2003. Managing and Conserving Soumhern African Grasslands Wimh High Endemism.
Mountain Research and Development 23:113–118. DOI: 10.1659/0276-
4741(2003)023[0113:MACSAG]2.0.CO;2.
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
SUPPORTING INFORMATION
Appendix S1. Scores allocamed mo mhree camegories based on personnel field observamion, each
weighmed according mo how DHM perceived a variable mo influence observabilimy. Condimions
were opmimal wimh a clear sky (score 1 = 100), cool memperamures (score 2 = 100) and calm
wind condimions (score 3 = 100). For omher weamher condimions, observabilimy was reduced and
we chose mhe scores according mo our subjecmive judgemenm of how much im affecmed our
abilimy mo demecm birds. For example, observabilimy was similarly reduced in smrong winds as in
hom weamher, emc. We mhen averaged mhe mhree scores mo gem a single value for observabilimy.
Appendix S2. Mulmi-Species, dynamic hierarchical model: R and BUGS code used mo fim mhe
model. R scripm wimh mhe JAGS model specificamion for mulmi-species hierarchical occupancy
model wimh effecm of grass heighm and grass cover on occupancy (Ψ), persismence (φ) and
colonisamion (γ) probabilimies wimh addimional effecm of environmenm (cloud cover and wind) on
demecmion probabilimy (p).
701
702
703
704
705
706
707
708
709
710
711
712
713
714
Figure 1(on next page)
Map of Ingula study area showing location of our random study plots created with
ggplot2
lon and lat on the y and x axis both represents longittude and latitude respectively
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
P01 P03
P03A
P07
P10
P22
P38
P39
P40
P41
P42
P50
P54 P57
P63
P64 P65 P66
P70
−28
.30
−28
.25
−28
.20
−28
.15
29.4
529.5
029.5
529.6
029.6
5
lon
lat
Figure 2(on next page)
Multi-season plot occupancy by month-November to February for the top 12 most
common species covering the two austral summers (2010/11-2011/12).
Occupie
d p
lots
Nov Dec Jan Feb
05
10
15
20
African Pipit
Occupie
d p
lots
Nov Dec Jan Feb
05
10
15
20
Cape Longclaw
Occupie
d p
lots
Nov Dec Jan Feb
05
10
15
20
Wing−snapping Cisticola
Occupie
d p
lots
Nov Dec Jan Feb
05
10
15
20
Red−capped Lark
Occupie
d p
lots
Nov Dec Jan Feb
05
10
15
20
Zitting Cisticola
Occupie
d p
lots
Nov Dec Jan Feb
05
10
15
20
Yellow−breasted Pipit
Occupie
d p
lots
Nov Dec Jan Feb
05
10
15
20
Common Quail
Occupie
d p
lots
Nov Dec Jan Feb
05
10
15
20
Long−tailed Widowbird
Occupie
d p
lots
Nov Dec Jan Feb
05
10
15
20
African Quailfinch
Occupie
d p
lots
Nov Dec Jan Feb
05
10
15
20
Banded Martin
Occupie
d p
lots
Nov Dec Jan Feb
05
10
15
20
Ant−eating Chat
Occupie
d p
lots
Nov Dec Jan Feb
05
10
15
20
Eastern Long−billed Lark
Figure 3(on next page)
Hierarchical plot occupancy of the 12 most common species, showing influence of grass
height and cover on persistence and colonization of each species during the austral
summer survey for two years 2010/11-2011/12 at Ingula
10 20 30 40
0.0
0.2
0.4
0.6
0.8
1.0
Grass height
Occupancy p
robabili
ty
10 20 30 40
0.0
0.2
0.4
0.6
0.8
1.0
Grass height
Pers
iste
nce p
robabili
ty
10 20 30 40
0.0
0.2
0.4
0.6
0.8
1.0
Grass height
Colo
nis
ation p
robabili
ty
APCLcWsCRcL
ZCYbPCQLtW
AQfBMtnAECELL
3 4 5 6 7 8
0.0
0.2
0.4
0.6
0.8
1.0
Grass cover
Occupancy p
robabili
ty
3 4 5 6 7 8
0.0
0.2
0.4
0.6
0.8
1.0
Grass cover
Pers
iste
nce p
robabili
ty
3 4 5 6 7 8
0.0
0.2
0.4
0.6
0.8
1.0
Grass cover
Colo
nis
ation p
robabili
ty
Figure 4(on next page)
Bird species richness comapring Ingula farms that experinced little grazing comapred to
neighbouring farms which were mostly annually burned and intensively grazed with
cattle during two austral summer surveys (2010/11-2011/12). Twelve plots were
surveyed
November
02
46
810
12
●
December
02
46
810
12
Eskom Private
02
46
810
12
January
●
Eskom Private
02
46
810
12
February
Specie
s r
ichness