RAITT_NISL_PRAC_2007
UNIVERSITY OF THE WESTERN CAPE
SUBJECT: CONSERVATION FOR CLIMATE CHANGE COURSE
MARKS: 100
TIME: 5 DAYS
Course Convener: Jay Reeler
Student: Gwen R. Raitt
Student Number: None yet. Old Staff Number: 260233
GENERAL INSTRUCTIONS AND ACCEPTANCE OF CONDITIONS
All STEPS ARE COMPULSORY
PLEASE SUBMIT A WORKED UP DOCUMENT ON THE TEMPLATE PROVIDED ENSURING ALL RESULTING IMAGES HAVE LEGENDS AND FULL CAPTIONS WITHIN THE MS WORD® DOCUMENT AND USING THE STYLE SHEET PROVIDED (MARKS WILL BE TAKEN OFF IF THIS IS CORRUPTED UP TO 10% AND I RECOMMEND THAT YOU USE REVEAL FORMATTING TO HELP YOU).
PLEASE ALSO SUBMIT YOUR WORKED-UP TEMPLATE ELECTRONICALLY TO
PLEASE MAKE SURE YOUR ANSWER TEMPLATE HAS YOUR NAME AND STUDENT NUMBER ENTERED AND IS SAVED USING THE FOLLOWING FILE NAMING/SUBMITTING CONVENTION-
SURNAME_NISL_PRAC_2007 (SO IF I WAS TO SUBMIT IT WOULD BE
REELER_ NISL_PRAC_2007.DOC)
PLEASE USE THE FOLLOWING AS A HEADER FOR EMAILING THE ASSIGNMENT
SURNAME_ NISL_PRAC_2007
(SUBSTITUTING SURNAME FOR YOUR SURNAME)
10% WILL BE DEDUCTED IF YOU DO NOT FOLLOW THIS CONVENTION FOR ELECTRONIC SUBMISSION OF YOUR PRACTICAL
WHERE IS THE DATA?
File Names – R.exe
R_additional_packages.exe
Climate_Change_Course_Practical.exe
DATA DESCRIPTION
You’ll need to run each of these executables. Install the software in the default directories that these programmes specify.
R.exe – This will install the R statistical package that you will use for running the GAM model and evaluating its accuracy.
R_additional_packages.exe – This will install the additional package Hmisc into the R directory. This package contains a library that is needed in order to run the model.
Climate_Change_Course_Practical.exe – This installs all the data and additional files that you will use in the practical to the C:\Climate Change Practical Folder. When unzipped, this folder should have the following files:
a2_2050.dbf
AMD_DATA.dbf
Climate_Change_practical.doc
Climate_Change_Practical_Template.doc
countries.dbf
countries.sbn
countries.sbx
countries.shp
countries.shx
Functions.R
national_iucn1to6_poly.dbf
national_iucn1to6_poly.sbn
national_iucn1to6_poly.sbx
national_iucn1to6_poly.shp
national_iucn1to6_poly.shx
R 2.0.1 (shortcut)
SACIC_WMS_help.doc
Script.R
All the data for modelling the species are contained within the AMD_DATA.dbf. This database file comprises XY locational data for Central and Southern Africa, at a resolution of 10’ by 10’. For each point, data on the average ecoregion is specified, as well as presence data for 27 species of African mammals, and each of six environmental variables.
The environmental variables are:
Table 1.
VARIABLE / DESCRIPTIONEVTR0112 / Net evapotranspiration (mm)
GDD10_0112 / Growth days over 10ºC (number of days)
MTC / Mean temperature of the warmest month (ºC)
MTW / Mean temperature of the coldest month (ºC)
PREC0112 / Mean annual precipitation (mm)
TMEAN0112 / Mean annual temperature (ºC)
The species for which data are provided are detailed on the following page (Table 2).
RAITT_NISL_PRAC_2007
Table 2.
Countries in which species is found / Common name / Scientific name / Indicator in DBF fileALL / African weasel / Poecilogale albinucha / AFR_WEAS
ALL / Giraffe / Giraffa camelopardalis / GIRAFFE
ALL / Blue, sky or diademed monkey / Cercopithecus mitis / SKY_MONK
ALL / Tsessebe / Damaliscus lunatus / TSESSEBE
Botswana / Black-footed or small spotted cat / Felis nigripes / BLACK_FT_C
Burundi / L'Hoest's guenon (monkey) / Cercopithecus lhoesti / LHO_GUENO
DRC / Dryas guenon (monkey) / Cercopithecus dryas / DRYAS_GUEN
DRC / Aquatic or fishing genet / Osbornictis piscivora / FISH_GENET
Kenya / Grant's gazelle / Gazella granti / GRANT_GAZE
Kenya / Plain zebra / Equus grevyi / PLAIN_ZEBR
Kenya / Puku / Kobus vardonii / PUKU
Kenya/Moçambique/Tanzania / Suni / Neotragus moschatus / SUNI
Malawi / Bushy-tailed mongoose / Bdeogale crassicauda / BUSH_MONGO
Malawi / Natal red duiker / Cephalophus natalensis / NATAL_DUIK
Moçambique / Dusky or Peters’ short-snouted elephant-shrew / Elephantulus fuscus / DUSK_SHREW
Namibia / Small grey mongoose / Galerella flavescens / GREY_MONGO
Namibia / Mountain zebra / Equus zebra / MOUNT_ZEBR
Namibia/Zimbabwe / Jameson's red rockhare / Pronolagus randensis / JAMES_RHAR
RSA / Capegrysbok / Raphicerus melanotis / CAPE_GRYSB
RSA / Cape or small grey mongoose / Galerella pulverulenta / CAPE_MONGO
RSA / Natal red hare / Pronolagus crassicaudatus / NATAL_HARE
RSA / Vaal or grey rhebok / Pelea capreolus / VAAL_RHEBO
Tanzania / Abbott's duiker / Cephalophus spadix / ABBOT_DUIK
Tanzania / Black-and-rufous elephant-shrew / Rhynchocyon petersi / BL_RUF_SHR
Tanzania / Harvey's red duiker / Cephalophus harvey / HARV_DUIKR
Uganda / Short-nosed or dusky-footed elephant-shrew / Elephantulus fuscipes / SH_N_SHREW
Zimbabwe / Nyala / Tragelaphus angasii / NYALA
RAITT_NISL_PRAC_2007
ANSWER 1- [10 marks]
Question 1Verify that the species you are planning to model does in fact fall within the limits of your selected country. Show the distribution of the species within your study country by changing the legend, (display the presence values over the countries shapefile) and discuss whether the distribution is suitable for modelling from the selected country. If it is not, specify any additional or alternative areas that you will include in the modelling process to ensure an appropriate model.
Use “Print Screen” and paste the screen capture into Windows Paint® to capture the images from ArcView® but save them in an image directory as *.jpg images. From this image directory you should paste the images into your MS Word® document (this will prevent accidental loss of images in the final document).
Legend
Figure. 1a Distribution of Harvey’s red duiker (Cephalophus harveyi) in Tanzaniaas per instructions (Resolution: 96x96 dots per inch).
Legend
Figure. 1b Distribution of Harvey’s red duiker (Cephalophus harveyi) in Tanzania showing the country’s borders (Resolution: 96x96 dots per inch).
Table 1. Quantitative Presence/Absence data for Harvey’s red duiker (Cephalophus harveyi) in Tanzania(Units: Map Cells)
Harvey’s Red Duiker / Number of cells of TanzaniaPresent / 761
Absent / 2007
Total / 2768
Discussion of the results from the above table and image (ensure it is unbolded and normal text)
Harvey’s red duiker (Cephalophus harveyi) is present in 27.49% of the map cells in Tanzania. This distribution seems suitable for modelling as it is neither completely absent nor everywhere present.
ANSWER 2- [15 marks]
Question 2Write a short description of the species you are modelling, including as many details of the ecology of the species (habitat, diet, breeding habits, rarity) as you can. Include a photograph of a specimen of the species (attribute your sources of information).
Figure. 2 Member of the genus Cephalophus(C. natalensis) to give a general idea of the modelled species’ appearance. I was unable to find a picture of Cephalophus harveyi on the web.
Source:
Describe the distribution and ecology of the species
Harvey’s red duiker (Cephalophus harveyi)has black legs and a black face. The rest of the coat is chestnut in colour. The average weight is about 15 kg. The shoulder height of Harvey’s red duiker is about 40 cm.[1] The African Mammal Databank indicates that this species was formerly part of Cephalophus natalensis (red forest duiker) and that its classification may need reconsidering. It occurs Tanzania (primarily), Kenya and Somalia. It might also occur in Ethiopia.[1, 2] The species IUCN status is Lower Risk: Conservation Dependent. Harvey’s red duiker occurs in subtropical and tropical dry forests (lowland,mountainand riverine) and forest mosaics or shrubland.[1, 2, 3] It feeds on fruit, leaves, twigs but also takes birds’ eggs, insects and carrion.[1]
References
1. Wikipedia contributors. Harvey's Duiker [Internet]. Wikipedia, The Free Encyclopedia; Updated 2007 Jun 15, 06:20 UTC [cited 2007 Jun 28]. Available from:
2. Anonymous. Cephalophus harveyi [Internet]. Artiodactyla, Bovidae, African Mammal Databank; Update unknown [cited 2007 Jun. 26—29]. Available from:
3. Antelope Specialist Group. Cephalophus harveyi[Internet]. Antelope Specialist Group, 2006 IUCN Red List of Threatened Species. IUCN 2006 Updated 1996 [cited 2007 Jun.28]. Available from:
ANSWER 3- [20 marks]
Question 3Specify for each of your three models the environmental characteristics you used in the GAM, their p-values, the overall p-value for the model, and the amount of the deviation in distribution the models explain. All of these values should be obtainable from the console once you have run a summary of the GAM. Explain whether these variables are direct or indirect descriptors of the distribution of your species, and what the implications of this might be. For the single variable model, explain why you chose your variable, and whether you think it is a good descriptor of the species’ distribution.
Table 2. Statistics for the single variable model.
Variable name / p-value / Total model Pr (>|z|) / Deviance explainedGDD10_0112 / < 2 x 10-16 / < 2 x 10-16 / 10.3%
Explain why you used this variable, and whether you think it is a good descriptor of the species distribution.
The variable ‘growth days over 10oC’ had shared the highest score for deviance explained with mean annual temperature. The variable is not a good descriptor of the species. Temperature acts both directly on the species and indirectly on the species’ food plants.
Table 3. Statistics for the 3 variable model.
Variable name / p-value / Total model Pr (>|z|) / Deviance explainedPREC0112 / 3.46 x 10-16 / 0.147 / 40.6%
TMEAN0112 / < 2 x 10-16
GDD10_0112 / < 2 x 10-16
Table 4. Statistics for the 5 variable model.
Variable name / p-value / Total model Pr (>|z|) / Deviance explainedMTW / < 2 x 10-16 / 0.243 / 49.1%
TMEAN0112 / < 2 x 10-16
GDD10_0112 / < 2 x 10-16
MTC / 0.0139
PREC0112 / 1.25 x 10-7
Explain whether the variables are direct determinants of the species distribution, and the implications of this for modelling:
All the temperature variables chosen are direct determinants of species’ presence. Temperature also affects species interactions[1]thus temperature also acts as an indirect determinant of species’ presence. Mean annual precipitation is only a direct determinant of the species distribution if the given species does not achieve all its moisture needs via its diet. The African Mammal Databank considers Harvey’s red duiker more likely to occur in suitable habitat within 1 km of standing water[2] so it is likely that mean annual precipitation is a direct determinant for Harvey’s red duiker. Mean annual precipitation is an indirect determinant of the presence of Harvey’s red duiker via biotic interactions. The model accuracy will be affected by the degree to which indirect determinants are correlated to the species of interest.
References
1. Graves J, Reavey D. 1996. Global Environmental Change: Plants, Animals & Communities. Harlow: Longman. 226p. 0-582-21873-X ISBN.
2. Anonymous. Cephalophus harveyi [Internet]. Artiodactyla, Bovidae, African Mammal Databank; Update unknown [cited 2007 Jun. 26—29]. Available from:
ANSWER 4- [5 marks]
Question 4Using one of the variables that is used in at least two of your models, show the plots of its response curves in each model. These should be screenshots from R. If you do not have the response curves available, you can run the command again from the relevant script, in order to capture the curves. Explain why these curves are different in different models, and why the margin of error (dotted lines) is higher at the edges of the distribution than in the middle.
Figure. 3a Plot of the response curve for ‘growth day over 10oC’ from the single variable model.
Figure. 3b Plot of the response curve for ‘growth day over 10oC’ from the three variable model.
Figure. 3c Plot of the response curve for ‘growth day over 10oC’ from the five variable model.
Interpret the differences between the curves, and explain why the margin of error is greater at the edges of the distribution:
Different variables are correlated with each other which explains the differences in the response curves. Other variables (e.g. biotic) have impacts at the edges of the distributionsso climatic variables will not be as accurate in determining the occurrence of the species there.[1] The climatic variables may be used to determine the fundamental niche of the species. However, the fundamental niche is likely to differ from the realised niche – the portion of the fundamental niche that the species has been able to occupy. This means that the climatic variables have a larger margin of error at the edges of the distribution.
Reference
1. Graves J, Reavey D. 1996. Global Environmental Change: Plants, Animals & Communities. Harlow: Longman. 226p. 0-582-21873-X ISBN.
ANSWER 5- [12 Marks]
Question 5By comparing the modelled and actual distributions of your species, as well as by comparing the receiver operating characteristic values of the respective models, describe how you decided on the most appropriate model to use for the determination of the future niche for your species. You should display the modelled distributions in ArcView using each of the models, as well as the actual distribution.
Table 5. Area Under the Curve (AUC) values from all three models for evaluation.
Number of variables used in model / AUC (calibration dataset) / AUC (evaluation dataset) / AUC (total)1 / 0.704504 / 0.6940888 / 0.7009262
3 / 0.9000161 / 0.905485 / 0.9016668
5 / 0.9312173 / 0.9339603 / 0.9321517
Legend
Figure. 4 The present distribution of Harvey’s red duiker (Cephalophus harveyi) in Tanzania(Resolution: 96x96 dots per inch).
Legend
Figure. 5 The distribution of Harvey’s red duiker (Cephalophus harveyi) in Tanzania as predicted by the single variable model (Resolution: 96x96 dots per inch).
Legend
Figure. 6 The distribution of Harvey’s red duiker (Cephalophus harveyi) in Tanzania as predicted by the 3 variable model (Resolution: 96x96 dots per inch).
Legend
Figure. 7 The distribution of Harvey’s red duiker (Cephalophus harveyi) in Tanzania as predicted by the 5 variable model (Resolution: 96x96 dots per inch).
Discuss the above results and differences, and explain which model you decided was best and why.
The single variable model yielded a visually poor match to the actual distribution of Cephalophus harveyi and its ROC values were less than 0.8. This was not surprising as only about 10.3% of the deviance was explained by the model. The single variable model was thus rejected. The 3 variable model showed some visual similarities to the species’ distribution and had a total ROC value of just over 0.9 so it is a good model but the 5 variable model had the best ROC value (> 0.93) and the best visual match to the actual distribution of Cephalophus harveyi so the 5 variable model was selected. The variables in the 3 and 5 variable models are correlated to each other.
ANSWER 6- [5 Marks]
Question 6What is the cutoff value determined by the function as appropriate for estimating species presence for your study species (using your final model)? This can be obtained from the R console window. Once again, if you cannot find it, rerun the relevant command from the script:
#Print out cutoff values in Console window
What, if anything, does this cutoff point imply about the model’s accuracy?
The 5 variable model had a probability cutoff of 0.2861414. The cutoff value is < 0.5 (equal chance of presence or absence) which suggests that the model may indicate that Harvey’s red duiker is present when it is not. The sensitive and specificity of the cutoffare 84.7569 and 84.75336 respectively. This suggests that the model has good precision in implementing the cutoff.
ANSWER 7- [20 Marks]
Question 7 Describe the likely impact of the climate change induced transformation of habitat for the conservation of this species. Where possible, combine your outputs with the IUCNNational Parks and designated conservation areas (national_IUCN1to6_poly.shp) shapefile to illustrate possible complications in terms of conservation areas. Using the attributes table for the IUCN parks shapefile, calculate the extent of conservation areas (in ha) in both present and future scenarios in which the species can be found. This value can be found in the “Gis_ha” column of the database for each park that falls within the distribution area.
Legend
Figure. 8 The future niche predicted by the 5 variable model for Harvey’s red duiker (Cephalophus harveyi) for the whole of Africa(Resolution: 96x96 dots per inch).
Legend
Figure. 9 The present distribution for Harvey’s red duiker (Cephalophus harveyi) for the whole of Africa(Resolution: 96x96 dots per inch).
Partial Legend
The reserves that match the distribution of Harvey’s red duiker (Cephalophus harveyi) are bright yellow.
Figure. 10 Map showing the existing reserves that presently protect Harvey’s red duiker (Cephalophus harveyi)(Resolution: 96x96 dots per inch).
Partial Legend
The reserves that match the 2050 projected climate envelope for Harvey’s red duiker (Cephalophus harveyi) are bright yellow.
Figure. 11 Map showing the existing reserves that overlap with the 2050 projected climate envelope for Harvey’s red duiker (Cephalophus harveyi)(Resolution: 96x96 dots per inch).
Discuss the implications for the conservation of your chosen mammal species under conditions of climate change by 2050 (under the A2 emissions scenario).
The projection suggests that the climate envelope for Harvey’s red duiker (Cephalophus harveyi) will shift north from its present distribution but still overlap the part of the present distribution. As this overlaps with the present distribution, it is likely that the habit will persist and possibly be able to shift north as well. Suitable climate also appears to the west and south of the present distribution – all the way south to South Africa in fact. This expansion of the area of suitable climate is, however, in the form of disjunct patches that are also disjunct from the species’ present distribution. This implies that, unless suitable climate bridges are present prior to 2050, Harvey’s red duiker would need translocation to reach these areas. The same disjunction, if not bridged prior to 2050, could prevent the duiker’s food plants from migrating and thus could mean that it would be unable to utilise the area even were it to reach the areas of suitable climate. At present, 2868050.84 ha of the distribution of Harvey’s red duiker is conserved. Within the projected 2050 distribution of climate suitable for Harvey’s red duiker, 59642667.42 ha are conserved. It thus appears that the conservation status of Harvey’s red duiker will remain as it is at present – conservation dependent. The threats of deforestation[1] are likely to remain and the duiker itself is likely to remain an addition to local menus[1] outside of conservation areas.