Universidade de Lisboa Faculdade de Ciências Departamento de Biologia Animal Factors ...

84 Pages·2013·3.96 MB·English

Universidade de Lisboa

Faculdade de Ciências

Departamento de Biologia Animal

Factors Shaping Bat Occurrence in Urban Green Areas

Gonçalo Filipe Fernandes Duarte

Dissertação
Mestrado em Biologia da Conservação

2013
Universidade de Lisboa

Faculdade de Ciências

Departamento de Biologia Animal

Factors Shaping Bat Occurrence in Urban Green Areas

Gonçalo Filipe Fernandes Duarte

Dissertação

Mestrado em Biologia da Conservação

Orientadores: Professor Doutor Jorge Palmeirim

Doutor Hugo Rebelo

2013
Factors Shaping Bat Occurrence in
Urban Green Areas

Faculdade de Ciências da Universidade de Lisboa
Dissertação de Mestrado em Biologia da Conservação

Gonçalo Filipe Fernandes Duarte

Orientadores: Professor Doutor Jorge Palmeirim
Doutor Hugo Rebelo

2013
Table of contents
Table of contents ................................................................................................................................... i
List of figures and tables ..................................................................................................................... iii
Acknowledgments ............................................................................................................................... vi
1. Abstract ........................................................................................................................................... 1
2. Sumário ........................................................................................................................................... 2
3. Introduction ..................................................................................................................................... 5
3.1. Biodiversity conservation and ecosystem services .................................................................. 5
3.2. Urban biodiversity .................................................................................................................... 6
3.3. Bats, biodiversity and urban areas............................................................................................ 9
4. Objectives ...................................................................................................................................... 12
5. Materials and methods .................................................................................................................. 13
5.1. Study area ............................................................................................................................... 13
5.2. Sampling design and bat survey ............................................................................................. 15
5.1. Vegetation assessment methods (adapted from Silva (2013)) .............................................. 16
5.1. Environmental/landscape Data ............................................................................................... 16
5.2. Statistical analyses.................................................................................................................. 20
5.2.1. Generalized linear mixed modelling ............................................................................... 20
5.2.2. Procedure ......................................................................................................................... 21
5.2.2.1. Correlogram ............................................................................................................. 21
5.2.2.2. Data exploration ....................................................................................................... 22
5.2.2.3. Modelling ................................................................................................................. 23
5.2.2.4. Model predictions .................................................................................................... 24
6. Results ........................................................................................................................................... 25
6.1. Spatial autocorrelation............................................................................................................ 27
6.2. Assessment of sampling conditions ....................................................................................... 28
i
6.2.1. Exploratory analysis and data filtering ........................................................................... 28
6.2.2. Sampling variables and candidate models selection ....................................................... 29
6.2.3. Final sampling model ...................................................................................................... 29
6.2.4. Sampling model adequacy .............................................................................................. 30
6.2.5. Sampling model predictions ............................................................................................ 30
6.3. Assessment of factors influencing bat occurrence ................................................................. 30
6.3.1. Exploratory analysis ........................................................................................................ 30
6.3.2. Variable and candidate models selection ........................................................................ 33
6.3.3. Outliers of selected explanatory variables ...................................................................... 33
6.3.4. Candidate and final models ............................................................................................. 33
6.3.5. Relative importance of variables ..................................................................................... 39
6.3.6. Model adequacy .............................................................................................................. 39
6.3.7. Model predictions ............................................................................................................ 39
7. Discussion ..................................................................................................................................... 46
7.1. Bat diversity ........................................................................................................................... 46
7.2. Sampling conditions ............................................................................................................... 48
7.3. Features promoting bat activity in urban green areas ............................................................ 49
7.4. Caveats and limitations .......................................................................................................... 52
7.5. Conservation implications ...................................................................................................... 53
7.6. Future studies ......................................................................................................................... 55
8. References ..................................................................................................................................... 56
Appendices ......................................................................................................................................... 74

ii
List of figures and tables
Figure 1 – Location of the study area. Example of one sampling scheme for one of the selected
urban green areas and their respective 20-meter exclusion zone. ...................................................... 14
Figure 2 – Example of bat calls analysis with spectrogram, oscillogram and power spectrum in
BatSound Pro v3.31b. ......................................................................................................................... 17
Table 1 – Description of climate and sampling-related variables. ..................................................... 17
Table 2 – Classification of vegetation clusters. ................................................................................. 17
Table 3 – Description of landscape and green area variables. ........................................................... 18
Figure 3 – Proportion of identification groups of bat-passes. .............................................................. 25
Figure 4 – Number of bat passes per urban green area. ....................................................................... 26
Figure 5 – Minimal number of species per urban green area. ............................................................. 26
Figure 6 – Number of bat passes from Pipistrellus species per average maximum energy frequency
(FmaxE). ............................................................................................................................................. 27
Figure 7 – Correlogram of the bat-passes with 95% pointwise bootstrap confidence intervals and
maximum lag distance of 11.2 km ..................................................................................................... 27
Figure 8 – Cleveland dot plot of the dependent variable (bat passes) and the explanatory variable for
the sampling condition analysis. Outliers marked with circles. ........................................................... 28
Table 4 – Univariate GLMM models. Summary showing the log-likelihood, Akaike information
criteria (AIC), Akaike differences (∆AIC), variable slope estimate (Estimate), variable slope
estimate standard error (Std. Error Estimate), and the lower and upper limits of the 95% confidence
interval for the variable slope estimate (in bold, those whose interval does not contain zero). See
Table 3 for variable abbreviations. .................................................................................................... 29
Table 5 – Sampling conditions assessment final models results. Summary showing the log-
likelihood, Akaike information criteria (AIC), Akaike differences (∆AIC), Akaike weights (AIC
weights), variable slope estimate (Estimate), variable slope estimate standard error (Std. Error
Estimate), and the lower and upper limits of the 95% confidence interval for the variable slope
estimate estimate (in bold, those whose interval does not contain zero). See Table 3 for variable
abbreviations. ..................................................................................................................................... 30
iii
Figure 9 – Plot of the Pearson residuals versus (A) fitted values; (B) temperature; (C) average wind
speed; (D) Green Area; (E) Month...................................................................................................... 31
Figure 10 – Plot of model prediction versus observed values of (A) temperature; and (B) average wind
speed. .................................................................................................................................................. 32
Figure 11 – Cleveland dot plot of the dependent variable, bat passes. Outliers marked with circles. .. 32
Table 6 – Groups of correlated variables with the Akaike information criteria (AIC) and Akaike
differences (∆AIC) results for univariate GLMM models. In bold the variables that are kept in the
analysis. See Table 3 for variable abbreviations. ............................................................................... 34
Table 7 – Variance inflation factors of the 19 variables. See Table 3 for variable abbreviations. .... 34
Table 8 – Univariate GLMM models results. Summary showing the log-likelihood, Akaike
information criteria (AIC), Akaike differences (∆AIC), variable slope estimate (Estimate), variable
slope estimate standard error (Std. Error Estimate), and the lower and upper limits of the 95%
confidence interval for the slope estimate (in bold, those whose interval does not contain zero). See
Table 3 for variable abbreviations. .................................................................................................... 35
Figure 12 – Cleveland dot plot of the variables that compose the candidate models. Outliers marked
with circles.......................................................................................................................................... 36
Table 9 – Results of the candidate models. Summary showing the log-likelihood, Akaike
information criteria (AIC), Akaike differences (∆AIC), Akaike weights (AIC weights). See Table 3
for variable abbreviations................................................................................................................... 37
Table 10 – Results of the final models. Summary showing additionally the Accumulated Akaike
weights (Accumulated AIC weights); See Table 9 for the remaining column abbreviation. See Table 3
for variable abbreviations................................................................................................................... 38
Table 11 – Number of final models were each of the variables are present and their relative importance
base on the sum of the Akaike weights. See Table 3 for variable abbreviations. ................................ 39
Figure 13 – Plot of the Pearson residuals versus (A) fitted values; (B) Green Area; (C) Month. Based
on model with the lowest AIC ............................................................................................................ 40
Figure 14 – Plots of the Pearson residuals versus observed values of (A) canopy perimeter; (B)
distance to forest edge; (C) relative canopy area in green area; (D) area of low shrubs. Based on
model with the lowest AIC ................................................................................................................. 41
iv
Figure 15 – Plot of the Pearson residuals versus observed values of area of resinous trees. Based on
the model with the second lowest AIC. ............................................................................................... 42
Figure 16 – Plot of model prediction versus the observed values of (A) canopy perimeter, (B) distance
to forest edge, (C) relative canopy area in green area and (D) low shrubby undercover area. ............. 43
Figure 17 – Plot of model prediction versus area of resinous forest observed values. ......................... 43
Figure 18 – Spatial predictions based on the 12 final models and their respective AIC weight. ...... 44
Figure 19 – Histogram of averaged predicted values per colour class and per green area present in the
spatial prediction map. Colours in accordance with those of Figure 18. Labels correspondance: A –
Alto dos Gaios Park, B – Costa da Guia Garden, C – Casino Garden / Sto. António Woodland, D –
Joaquim Ereira Woodland / Rotários Pinewood / R. dos Mochos Urban Park, E – Garden of the
Museum of Portuguese Music, F – Green Area of Outeiro da Vela, G – Outeiro de Polima Urban
Park, H – Green Area of Outeiro dos Cucos, I – Marechal Carmona Park, J – Palmela Park, K –
Junqueiro Pinewood, L – Ingleses Pinewood, M – Quinta da Alagoa Park, N – Quinta de Rana Park,
O – Quinta de S. Gonçalo Park. ......................................................................................................... 45
Figure 20 – Diagram illustrating the creation of an urban green area with the adequate features to
promote bat activity. ........................................................................................................................... 54
Table I – Noise scale used during sampling. ..................................................................................... 74
Table II – Example of the commands used in R.................................................................................. 75

v
Acknowledgments
Primeiro gostaria de agradecer ao Hugo Rebelo por ter acedido ao meu pedido de ajuda para este trabalho, mesmo
quando apenas era um projecto de intenções enfiado na minha cabeça. Foi graças ao seu entusiasmo, durante uma
formação com ele, que começou a aventura pelo mundo dos morcegos. Este trabalho não teria sido possível sem a sua
paciência e empenhada colaboração.
Um profundo agradecimento ao Professor Jorge Palmeirim por ter aceite orientar-me, pelos conselhos, revisões e
preciosas dicas que muito contribuíram para melhorar este empreendimento.
Sem o incentivo e apoio da Câmara Municipal de Cascais, particularmente através da Cascais Ambiente, este trabalho
não teria sido possível de concretizar-se. Aos colaboradores da Cascais Ambiente (Alexandre, Andreia, Bernardo,
Bruno, Inês, Irene, José, Margarida, Sara, Vítor e Vasco) que ajudaram na amostragem nocturna o meu obrigado. Um
agradecimento em especial ao Vasco pela ajuda com a parte da vegetação.
Gostaria de agradecer à Sara Saraiva da Cascais Ambiente, cujo empenho foi muito para além daquilo que o seu
profissionalismo lhe exigia. A sua ajuda e dedicação foram decisivas para que este trabalho se concretizasse. Pelas
incontáveis horas de amostragem nocturna (a grande paciência nessas horas) e pelos diversos inputs ao longo do
trabalho um enormíssimo obrigado.
Um gigante agradecimento ao Fernandinho pela companhia na amostragem nocturna, sem ti mais de 1 terço do trabalho
teria de ser feito a solo! Já agora um pedido de desculpa pelos enjoos nas viagens de carro! Felicidades e espero que
tudo corra bem por terras norueguesas.
Ainda, e finalmente em relação à amostragem nocturna, um obrigado àqueles cuja curiosidade ou apenas vontade de
ajudar os levou a colaborarem: Paulo Gonçalves, José Cáceres e Tio Rui.
Aos amigos Paulo e Sofia um obrigado pela participação na amostragem nocturna, pela paciência com a "pedinchice"
de artigos, mas sobretudo um valente obrigado pelos debates estatísticos, dicas e revisões preciosas.
Um grande abraço de profundo agradecimento ao Teixeira pelas sugestões e revisões anglo-saxónicas. Muitos parabéns
pelo teu rebento.
Aos meus avós Dilene e Alberto gostaria de lhes agradecer todas as palavras de carinho e encorajamento pois, apesar de
não perceberem porque raio andava eu à noite "à caça" de morcegos, sempre me apoiaram.
Aos meus pais qualquer agradecimento será parco! Obrigado por terem dedicado a vossa vida a tentar que os vossos
filhos tenham todas as oportunidades de serem felizes. Finalmente, obrigado por terem tido a paciência de ir à noite ver
um tonto a olhar para o ar com umas maquinetas na mão.
À Margarida um obrigado pela ajuda na amostragem nocturna, pelas discussões de estatística, dicas e revisões. Obrigado
por seres quem és, por toda a paciência e apoio que me tens dado, sobretudo nestes "simpáticos" dois últimos anos.
Finalmente, gostaria de dedicar este trabalho ao meu sobrinho Alexandre (mana um grande obrigado pelo teu filho!).
Este rapazola tem "o sorriso"... daqueles que tudo relativiza e nos anima incondicionalmente!
vi
1.Abstract
Urbanisation is one of mankind’s longer-lasting activities. Bat species most sensitive to human
activities have suffered population declines due to urbanisation and the resulting loss of habitats.
The presence of urban green areas in a city can promote the presence of bats and increase their
activity in general. Little is known about the characteristics that an urban green area should have to
promote the presence of bat species. This study was conducted in the municipality of Cascais and
addresses three questions: What features of urban green areas promote bat activity? Which urban
green areas in Cascais have best conditions for bat occurrence? Which urban green areas should be
improved? During 67 nights, between September 2011 and October 2012, 354 points in 15 urban
green areas were sampled with bat detectors. A total of 39 variables were used to describe the urban
green areas, and their surroundings. The data were statistically analysed using Generalized Linear
Mixed Models. The most relevant variables were canopy perimeter, proportion of canopy area in
green area, and area covered by low shrubs. Higher bat activity occurs where canopy perimeter is
greater, when the relative amount of canopy in a green area is high and where shrub vegetation
occurs. Bat activity tends to be null in places where the distance to forest edge is greater than 50m.
Four urban green areas stand out by having a high number of predicted bat passes, while in two
urban green areas predictions indicate very low levels of bat activity. Results suggest that in order
to promote bat activity, urban green areas should be forested, have clearings, extensive edge
perimeters and low shrubby vegetation. The majority of urban green areas in the eastern part of the
municipality need improvement in order to promote bat presence.
Keywords: Bats, Urban Green Areas, Urbanisation, Cascais
1