ENMTools User Manual
Contents
1. e bd o pi x o e o ee o ec gt e m e s Actual Replicate 1 In this example we have two species red and green that are separated by a linear boundary and we wish to test the hypothesis that this geographic boundary between these species corresponds with a significant environmental boundary with significance defined as more different than expected by chance To test this hypothesis ENMTools randomizes the location of the boundary itself ENMTools does this by pooling the occurrences for the two species randomly selecting a slope for the linear boundary and then choosing an intercept such that the new boundary splits the pooled occurrences into artificial species with the same sample sizes as in our original data set in this case 18 green and 15 red ENMTools then construct ENMs for the two newly generated species and measures overlap between them By doing this many times ENMTools is able to construct a distribution of expected overlaps across a randomly drawn linear barrier that splits populations into the sample sizes present in our actual data By comparing the niche overlap measurement from the empirical data to those obtained after randomly drawning the barrier we can ask whether the real barrier is partitioning habitat into more environmentally distinct regions than expected by chance There are a few things to be cautious about with the linear range breaking analysis F2. r A a s o o e m n o Eora A i e A e oS i 4 e e e o r o ee io e j e e E n n e e Actual Replicate 1 In this example we have two populations that are allopatric but the barrier between them does not appear to be linear Instead of comparing the overlap between them to a distribution of overlaps produced by randomly drawn linear barriers we will compare them to a distribution of overlaps produced by randomly drawing pseudoreplicate populations with non linear ranges a k a blobs As with linear range breaking the analysis begins by pooling the occurrence records for the two populations of interest We then randomly select a single occurrence point labeled start in the figure above From this starting point we continue adding the nearest occurrence points until we construct a data set of the same size as the less numerous of our two species This effectively draws the smallest possible polygon around our starting point that includes the required number of points The remainder of the comparison proceeds in the same manner as the linear analysis 19 Although the blob method of generating spatial partitions works well in situations where ranges are elongate and sample sizes are asymmetric having a large number of repeated occurrences can still lead to reduced independence of pseudoreplicates Again we recommend deletion of duplicate pseudoreplicates
3. Levins R 1968 Evolution In Changing Environments Monographs in Population Biology volume 2 Princeton University Press Princeton New Jersey USA Phillips S J R P Anderson and R E Schapire 2006 Maximum entropy modeling of species geographic distributions Ecological Modelling 190 231 259 Phillips S J 2008 Transferability sample selection bias and background data in presence only modelling a response to Peterson et al 2007 Ecography 31 272 278 Schoener T W 1968 Anolis lizards of Bimini resource partitioning in a complex fauna Ecology 49 704 726 Warren D L R E Glor and M Turelli 2008 Environmental niche equivalency versus conservatism quantitative approaches to niche evolution Evolution 62 2868 2883 26
4. Testing a wide range of comparative hypothesis is ENMTools s raison d etre Hypotheses that may be tested with ENMTools include 1 The niche identity test This test is used to ask whether ENMs generated from two or more species are more different than expected if they are drawn from the same underlying distribution 2 The background similarity test This test is used to ask whether ENMs drawn from populations with partially or entirely non overlapping distributions are any more different from one another than expected by random chance 3 Range breaking tests This suite of tests is used to ask whether biogeographic boundaries correspond with significant environmental variation ll c i Identity tests 11 gwa O ee ENM measurements Hypothesis testing Resampling Options Test the hypothesis that two populations were drawn from an identical distribution of environmental variables Occurrence files Addfiles Import file list _ Save file list Clear file list Number of replicates v Run Maxent Keep pseudoreplicate files Binary predictions using minimum training presence GO The niche identity test allows the user to test whether the habitat suitability scores generated by ENM models from two species exhibit statistically significant ecological differences It does this by pooling empirical occurrence points and randomizing their id
5. file list Clear file list Number of replicates Proportion of records to delete d n Run Maxent At present ENMTools offers four methods of randomly resampling data 1 nonparametric bootstrap 2 delete D jackknife 3 delete 1 jackknife k fold cross validation and 4 retain X jackknife 1 When these resampling analyses were initially written into ENMTools Maxent didn t allow resampling via its own GUI Now that Maxent offers similar functionality the resampling tools in ENMTools may be largely obsolete before they are ever used publicly most users will likely prefer to conduct their resampling in Maxent due to the availability of confidence intervals on marginal suitability functions Nevertheless ENMTools s resampling functions may be useful in some instances particularly when resampled data sets for use with other ENM construction methods than Maxent are desired 22 1 Nonparametric bootstrap This method creates replicates from the original data set by resampling with replacement The user simply selects a list of files and the number of replicates to execute 2 Delete D jackknife This method builds replicate data sets by randomly deleting a portion of the data specified by D and constructs ENMs using the remainder For this analysis the user selects a list of files to analyze the number of replicates to construct and analyze and the portion of records 0 lt D lt 1 to delete ea
6. input files could then be analyzed using other methods of ENM construction Keep pseudoreplicate files The button marked keep pseudoreplicate files tells ENMTools whether or not it should delete the files Maxent generates for each pseudoreplicate ENM When you first begin analyses of a given dataset it will be useful to individually inspect results of each pseudoreplicate to ensure proper behavior of ENMTools and Maxent However saving all of the data associated with each pseudoreplicate for a large analysis can quickly eat up all your hard drive space Binary predictions using minimum training presence This button is used to conduct analyses on simple binary predictions that involve predicted presence or absence rather than some quantitative measure of habitat suitability Because Maxent produces quantitative measure of suitability the use of a binary presence absence prediction requires simplification of the original output from Maxent The simplest approach to doing this is to assign a threshold value to the Maxent predicted suitability output scores that corresponds with the minimum training presence i e the lowest predicted suitability score that corresponds with a known occurrence When this option is used in ENMTools the identity test is conducted on grids that consist strictly of Os species assumed to be absent due to predicted habitat 13 suitability score lower than the minimum training theshold and 1s species assumed to be
7. present due to predicted habitat suitability score higher than minimum training threshold Some caution should be applied in interpreting the results from such binary predictions in addition to the general concerns expressed about comparing thresholded niche models in appendix S1 of Warren et al 2008 there are potentially conceptual problems with comparing the overlap between thresholded models for two species to the overlap seen between thresholded models for subsets of their pooled occurrences One word of caution be sure that you measure your actual overlap using ENMs that were built using the same data and same suitability measure as your pseudoreplicates Comparing D to I makes little sense and comparing overlap of logistic Maxent ENMss to overlap of raw or cumulative ENMs makes even less 14 ll c ii Background tests ENM measurements Hypothesis testing Resampling Options Test the hypothesis that two species are more or less similar than expected based on some estimate of the environmental background from which they are drawn Focal species occurrend Not set Background samples Not set Number of replicates Number of background samples Scheduled analyses Add this analysis Cleartist e V Run Maxent Keep pseudoreplicate files Binary predictions using minimum training presence GO This test uses randomization to determine whether two species are more or less similar than
8. yes Perl code no damn way the right to show a ll packages C Acme Beatnik 0 02 Source Filter to implement the Beatnik language C Acme BeyondPerl 0 01 RDBMS calculates instead of Perl 3 Acme BlahBlahBlah 0 01 Perl extension For blah blah blah Scroll down u ntil you fi nd the Acme Bleach 1 12 Perl programming in morse code package Tk and then right click on Acme Bleach Nu 0 04 Fit the whole world between 0 and 1 io it to install Status Details Synchronizing Database Downloading ActiveState Package Repository packlist not modified Downloading CPAN packlist done Synchronizing Database done 6539 packages 6539 listed 37 installed O to install O to remove Install Area site I c Launching ENMTools ENMTools can be launched in several ways If you are using one of the binary versions it can be launched simply by double clicking on the icon for the executable file The Perl version can be launched this way as well if your system is configured so that Perl is the default application for files with the pl extension If this is the case you can also launch ENMTools from the command prompt by typing the appropriate path e g ENMTools pl if you are currently in the directory with the perl script If your system isn t configured to open pl files with Perl you can launch ENMTools by opening a command prompt and typing lt perlpath gt lt enmtoolspath gt replacing each of those with the appropria
9. ENMTools User Manual v1 0 Dan Warren Rich Glor and Michael Turelli danwarren ucdavis edu I Installation a Installing Perl b Installing Tk c Launching ENMTools Il Running ENMTools a The options menu i ENMTools options ii Maxent options b The ENM Measurements menu i Measuring niche overlap ii Measuring niche breadth c The hypothesis testing menu i Identity tests ii Background tests iii Range breaking tests d The resampling menu i Jackknife bootstrap ii Spation cross validation Ill Citing ENMTools IV Literature cited in this manual Thanks for using ENMTools At present this software should be considered an alpha release some bits aren t meant to be used yet and it hasn t been extensively user proofed yet It s still entirely possible to tell the software to do things that it shouldn t do which will produce nonsense results I Installation l a Installing Perl ENMTools is a Perl script with a graphical user interface that is implemented via the Tk package This means that you need to have Perl and the Tk package installed before you can use it It is unlikely that the versions of Perl and Tk needed to run ENMTools are already installed on your computer First and foremost you need Perl If you don t have it go to http www activestate com Choose Language Distributions gt ActivePerl The Standard distribution has everything you need and is free Download it and install it in jus
10. ally overinflates estimates of variance in ENMs when sample sizes are small and D is large Warren and Phillips in prep Ideally all of these methods should produce identical estimates of the uncertainty in suitability scores that is due to incomplete sampling sampling variance However there are reasons why these methods may not be precisely equivalent when applied to ENMs The nonparametric bootstrap allows individual occurrence points to appear in a single pseudoreplicate multiple times Given that many ENM construction efforts begin with trimming duplicate occurrence points some users have expressed concern that re introducing these duplicates may produce unreliable estimates of variance If this is the case one of the jackknife methods may be preferable However jackknife methods necessarily produce pseudoreplicate data sets that are smaller than the original data set which may alter feature selection and other aspects of the modeling process This may result in features being selected or not during jackknife runs that might not be preferred if the full data set was used This could affect jackknife s ability to inform variable selection and may be partially responsible for inflated estimates of variance under the conditions mentioned above The delete 1 jackknife is unlikely to exhibit this problem to any great degree A systematic comparison of nonparametric bootstrap and delete D jackknife methods as applied to ENMs is currently under
11. ch time 3 Delete 1 jackknife Also known as k fold cross validation this method constructs N pseudoreplicates for a data set of size N each of which is missing one of the points in the original data set The user simply selects a list of files as the number of replicates is determined from the sample size 4 Retain X jackknife This method builds replicate data sets by deleting all but X occurrences where X is an integer Unlike the delete D jackknife this method will keep sample size for pseudoreplicates constant across species even if the sample sizes for the original data sets differ For example if X is set to 20 and you analyze two species one with a sample size of 100 and one with a sample size of 200 replicates for both species will be constructed using 20 points For this analysis the user selects a list of files to analyze the number of replicates to construct and analyze and the number of records 0 lt X lt N to retain each time All of these methods calculate summary statistics mean and variance for all runs once Maxent analyses are complete Due to autocorrelation between pseudoreplicates a variance inflation factor is applied to the delete D retain X and delete 1 jackknife so that the variance estimates will more accurately represent the true uncertainty in habitat suitability Efron and Tibshirani 1994 p 149 Although this procedure is standard for jackknife resampling there is reason to suspect that it gener
12. e desert habitat in the above example is 10 km wide we construct pseudoreplicates with a 10 km wide ribbon Because our hypothesis deals with the amount of environmental heterogeneity expected for a band of habitat of a certain size we keep the width of the barrier rather 20 than sample size constant This means that sample sizes for A B and C may change slightly for different pseudoreplicates so the ribbon analysis is best applied in situations where sample sizes for both species are high enough that minor deviations in sample size will have little effect on ENMs ENM measurements Hypothesis testing Resampling Options Randomly partition the pooled range of two species using a linear barrier of a specified width Files containing occurrences Add files Import file list Save file list Clear file list Number of replicates Width of barrier Run Maxent The ribbon range breaking page is identical to the linear and blob range breaking pages with the exception of the field width of barrier The width provided here should be in the same units as the cellsize argument in your ASCII environmental layer files e g if your ASCII files are in arc seconds the barrier width should be also 21 Il d Resampling ll d i Jackknife Bootstrap Beno Oe ENM measurements Hypothesis testing Resampling Options Delete d jackknife Files containing occurrences Add files Import file list Save
13. entities to product two new samples with the same numbers of observations as the empirical data See Warren et al 2008 for details To the niche identity test you need to use the Add Import Save and Clear buttons to select a set of csv occurrence files for analysis NOTE unlike analyses of ENM similarity the files used for hypotheses testing are csv files of occurrence points not asc files of habitat suitability scores Once the files with the desired occurrence points have been selected the program then conduct pairwise identity tests for every pair of occurrence points Although we ve found that it is often easier to simply maintain different csv occurrence files for each species or population you intend to analyze hypothesis testing in ENMTools can also accommodate a single csv with multiple different labels for occurrence points If a file with multiple different occurrence labels is used the points with the different labels will be treated as distinct datasets and every possible pairwise comparison will be conducted by default e g if you have one file with species A and B and another with species C and D you will be doing comparisons A C A D B C and B D as well as A B and C D 12 Hypothesis testing is conducted by default on the set of layers that was previously Identity Test specified in the Options tab It is important to note that the test of niche identity and 60 other hypothesis tests conducted in ENMTools can ta
14. eps to configure the program before proceeding any further This is necessary because ENMTools needs to know a few things about where your data and other files are before it can do anything trust us this required configuration stage will save you time down the road If it is not properly configured ENMTools will confess its ignorance at start up by complaining about a missing config file To configure ENMTools you need to go to select the options tab from the ENMTools menu 1 Tell ENMTools where your environmental data can be found There are two possible locations for your environmental data 1 embedded directly in your list of occurrence points species with data or 2 as a separate file or set of files climate layers If you select the latter more frequently used option you will then need to tell ENMTools where your climate layers are located by clicking on the Layers directory button and navigating to the appropriate folder once this step is completed the directory listed to the right of the Layers directory button should correspond with the location of your layers 2 Specify an output directory for your results ENMTools also needs an output directory where it can create manipulate and analyze files It s a VERY good idea to generate and specify a separate directory for each analysis you conduct some ENMTools analyses generate lots of output files and keeping track of these files can quickly become
15. expected based on the differences in the environmental background in which they occur The identity test above is a very strict condition for assessing ecological similarity identity is expected to be rejected only when species tolerate the exact same set of environmental conditions and have the same suite of environmental conditions available to them Although the former condition may hold for some species pairs the latter is unlikely to hold for allopatric species To conduct each background test you need two files The first file is a csv file with occurrences for a single species the focal species for this particular test The second file is either a csv file of points lat long from the designated background region or an ASCII raster file to use as a mask to generate random background points If a mask file is provided any cell designated 9999 is interpreted as being outside the study area and any other value is interpreted as being inside In either case generating the background file requires a bit of knowledge of a GIS program like ArcGIS Perhaps the simplest approach is to draw a polygon around the background area in ArcMap and export the region inside this polygon as a new layer that will be used as the mask in ENMTools 15 As an example say you want to compare a pair of species A and B to see whether they were more or less similar than expected based on the environmental conditions available to them You wo
16. hat limits the amount of RAM that the Java virtual machine will allow you to allocate to Maxent As far as we ve been able to figure out right now this isn t an ENMTools problem or a Maxent problem it seems to be a Java problem If anyone knows of a good solution please contact Dan at dan warren gmail com we would appreciate the help Il b The ENM Measurements Menu Selecting the ENM measurements tab permits you to calculate two basic metrics from predicted habitat suitability scores generated by Maxent or some other niche modelling program 1 quantification of niche overlap between ENMs generated from two or more species and 2 assessment of niche breadth Conducting these analyses requires that ENMs and associated habitat suitability scores have already been generated by another software package Il b i Measuring niche overlap gwe i O ENM measurements Hypothesis testing Resampling Options Measure niche overlap between two or more ASCI raster files Files to be compared Save file list Clear file list Name for this analysis The overlap page is for measuring similarity between predictions of habitat suitability between one or more pairs of populations using methods introduced by Warren et al 2008 The setup is fairly straightforward you can add files save and import lists of files and name your analysis The files to be added in this window should be ASCII files of predicted habitat su
17. irst randomly drawing linear barriers like this can become problematic when the pooled set of occurrences is much longer in one spatial direction than in others This is due to the limited number of ways there are to 18 bisect very elongate ranges in the extreme case where all occurrences fall along a single line there are at most two ways to split the pooled set of occurrences This clearly leads to a lack of independence of pseudoreplicates The same sort of problem can occur when sample sizes for the two populations are highly asymmetric when this is the case occurrence points near the center of the set of pooled ranges will preferentially be assigned to the species with the most occurrences in the original data set once again leading to a lack of independence of pseudoreplicates Finally having many replicate occurrence points for either of the two species i e when a larger portion of one or both of their occurrences are repeated records for the same geographic coordinates can also lead to a lack of independence of pseudoreplicates as it restricts the number of ways that the pooled occurrences can be bisected To avoid a lack of independence among pseudoreplicates it is necessary to carefully examine resulting output and delete duplicate pseudoreplicates i e pseudoreplicates that result in identical partitions Blob Range breaking The blob method of range breaking is in most respects similar to the linear analysis
18. itability produced by Maxent or some other method The program will automatically measure overlap using two different statistics Schoener s D Schoener 1968 and the statistic see Warren et al 2008 for additional details All you have to do to measure overlap is load in a list of asc files and provide a name for the summary output file Every pairwise comparison will be made so if you load in a list of ten files you re making 45 comparisons Execution time depends on the number of ASCII files and the resolution of the data The output of an overlap analysis is two files each named using the analysis name you provided These files can be opened using a text editor or spreadsheet program Each one contains a table of pairwise overlaps using the statistic listed in the file name Both and D range from 0 species have completely discordant ENMs to 1 species have identical ENMs Il b ii Measuring niche breadth Benn cts O O O O O O O O O OOOO O O ENM measurements Hypothesis testing Resampling Options Measure niche breadth on ASCI raster files Files to measure Add files Save file list Clear file list Name for this analysis This page works almost exactly like the niche overlap page except that it outputs a csv file containing measurements of niche breath for each of the asc files that you specify The two niche breadth metrics are those of Levins 1968 10 il c The Hypothesis Testing Menu
19. ke a long time depending 4 I PIS rw on the size and resolution of the spatial 30 data The output from the niche identity test includes a pair of csv files with niche io overlap scores from each pseudoreplicate These scores represent the expected SL PF oF LP 0 GF FH 07 LW OP LH 9 LW OP degree of niche overlap when samples are drawn from the same distribution i e the pooled sample of occurrence points from two populations By comparing the overlap between ENM models generated from the actual data for each species obtained using the overlap tab under the ENM measures menu to the null distribution obtained using the identity test tab it is possible to ask whether ENMs produced by two populations or species are Statistically significantly different In the example to the right the red arrow indicates the measured overlap between species and the histogram illustrates the distribution of overlaps from pseudoreplicates It is clear from this example that the ENMs built based on the actual occurrences of the two species are more different than expected by chance This outcome amounts to a rejection of the hypothesis of niche identity Options for running the Identity test Run Maxent By default the Run Maxent dialogue box should have a checkmark in it This simply means that By de selecting Run Maxent you can tell ENMTools to simply generate input files for pseudoreplicates but not analyze those files in Maxent These
20. phic regions are fairly clear Other situations are not as simple The results of this test can vary based on the definition of the background areas so users are strongly encouraged to either 1 have a clear biological justification for defining background regions or 2 conduct sensitivity tests by specifying alternative background regions These analyses can be very time consuming depending mostly on the size and resolution of the spatial data 16 ll c iii Range Breaking 52 ENMTools ENM measurements Hypothesis testing Resampling Options Randomly partition the pooled range of two species along a straight line Files containing occurrences Save file list Number of replicates Run Maxent These range breaking tests are used to ask whether geographic boundaries between species or populations are associated with significant environmental variation Glor and Warren in prep Range breaking tests come in three main flavors linear blob and ribbon The linear and blob tests are two versions of a test that permit one to ask whether the geographic regions occupied by two species are more environmentally different than expected by chance The ribbon test meanwhile is designed to test whether the ranges of two species are divided by a region that is relatively unsuitable to one or both forms To demonstrate what s happening with the range breaking tests take the following examples 17 Linear Range breaking
21. prior to statistical inference from blob range breaking analyses Ribbon Actual Partitioned Replicate 1 The ribbon range breaking analysis is used to test the hypothesis that two populations occurring in similar highly suitable habitats are separated by a band of less suitable habitat of a known width e g a band of desert habitat separating two forested regions In the Actual distribution in the above example the red species is primarily in area A and the green species primarily in area B Area Cis a region where the species geographic distributions overlap but is hypothesized to be marginal habitat for both In this analysis we pool the samples in the region of overlap and measure overlap between the two allopatric flanking regions A vs B and each flanking region and the ribbon of unsuitable habitat A vs Cand B vs C Finally we can measure the overlap between the pooled occurrences of the two species on the outside of the putative barrier and those on the inside A B vs C In order to estimate the distribution of expected overlaps for these four measures ENMTools first picks a slope and intercept that divide our pooled set of occurrences into samples of the size of the two species in our actual data just as with the linear range breaking test Following that we expand the linear barrier outwards until it is the same width as the band of putative unsuitable habitat that we are studying i e if th
22. seeing or not seeing Maxent s interface when ENMTools sends it data Viewing the interface can be useful when diagnosing problems but it may also slow down analyses somewhat You can choose whether to show the interface or not using the Show Maxent GUI radio button 7 Save your configuration Once you ve made all of your choices save your configuration using the save options button This creates a text file named ENMTools config which will automatically be loaded whenever you start ENMTools ll a ii Maxent Options gwe i o O ENM measurements Hypothesis testing Resampling Options Options for Maxent RAM to assign to Maxent MB mx1000m Make response curves for ENMs No Yes Make pictures for ENMs No Yes Make ROC plots for ENMs No Yes This page also under the options menu contains options that ENMTools will pass to Maxent for each run that it executes At the present time you are not able to manipulate the complete range of Maxent s options We will probably enable more options as time goes on All of currently implemented options should be fairly self explanatory with one exception the RAM to assign to Maxent option requires an argument of the form mx m where is the number of megabytes of RAM that you want Maxent to use Make sure that it doesn t exceed the amount of memory available on your system At present there is a bug on some 64 bit Windows systems t
23. t like a regular windows or Mac OSX program Where you install it doesn t matter to ENMTools but you ll want to make sure you can easily find this location later I b Installing Tk Once you ve got the newest Perl Package Manager Bax Activestate Perl distribution installed SE E t Yew Action Help 5 5 SIQI a a x ENMTools should run right away If it BIR Bos a e Package Name l Area Installed Available Abstract a doesn t you mayn eed the newest abbreviation 0 02 Perl pragma to abbreviate class names a f i i ABI 1 0 Perl module to parse chromatogram files generated b Tk distri buti on For thi S you need AcePerl 1 91 Object Oriented Access to ACEDB Databases to launch the Perl Package Manager PAH nni ACH perl abject g RNE ACH Builder 0 02 Tools for Building ACH Automated Clearing House F You can find this in the Start menu ACH Generator 0 01 Generates an ACH formatted file From an ACH perl o ACH Parser 0 01 Parse an ACH formatted File to ACH perl object folder that was created during the CD Acme 1 11111 The Base of Perfection Perl installation You can then either Acme 24 0 01 Your Favourite Tv show Acme module X D Acme 6502 0 0 6 Pure Perl 65C02 simulator press ctrl 1 or click the button C Acme Anything 0 02 Anything even imaginary modules are loadable highlighted in green in the image to D Acme BabyEater 0 04 Baby eating has never been easier C Acme BadExample 1 00 Perl document
24. te directory path e g usr bin perl ENMTools pl If ENMTools is failing to launch or exiting unexpectedly try launching it from a command line This won t fix the problem but will keep the console from automatically closing so that any error messages can be seen ll Running ENMTools ENMTools menubar has four basic options each of which provides you with a pull down menu and suite of associated options 1 ENM Measurements 2 Hypothesis Testing 3 Resampling and 4 Options ll a The Options Menu You will need to begin by familiarizing yourself with the Options menu because some basic features of ENMTools and an associated program Maxent need to be configured properly before you go any further ll a i Configuring ENMTools r g2 ENMTools ENM measurements Hypothesistesting Resampling Options Options specific to ENMTools Source of climate data D Species with data csv Climate layers Layers directory C Users danw Desktop Sample Data Sample Data Layers Output directory C Users danw Desktop Sample Data Sample Data output Projection directory Maxent Jjar file C Users danw Desktop maxent_3 3 1 jar Suitability measure Raw Logistic Cumulative Show Maxent GUI No Yes Maxent version 3 2 x or older 3 3 or newer Save options Once ENMTools has launched for the first time it is essential that you complete a few basic st
25. tes to perform The caveats mentioned in section Ill c about the possible non independence of pseudoreplicate data sets in certain situations also apply to these tests and must be taken into consideration when using them 25 lll Citing ENMTools We kindly request that users who publish results obtained from ENMTools cite the paper that introduced ENMTools s core methods Warren et al 2008 and the application note that formally introduces the program Warren et al In press Warren D L R E Glor and M Turelli 2008 Environmental niche equivalency versus conservatism quantitative approaches to niche evolution Evolution 62 2868 2883 Warren D L R E Glor and M Turelli In press Users who are analyzing data obtained from Maxent for ENM measurement analyses in ENMTools or using any of ENMTools s hypothesis testing functions should also cite the paper that introduces Maxent Phillips S J R P Anderson and R E Schapire 2006 Maximum entropy modeling of species geographic distributions Ecological Modelling 190 231 259 ENMTools users who make use of the range breaking functions are also asked to cite the paper that introduced these methods Glor and Warren In press Glor R E and D L Warren In Press Thanks in advance for using and citing ENMTools IV Literature Cited in this Manual Efron B and R J Tibshirani 1994 An Introduction to the Bootstrap Chapman amp Hall CRC New York USA
26. uld compare the species occurrences for A to the background for species B using either a set of points or an ASCII mask for the background area The number of replicates determines the statistical resolution of the test 100 replicates is enough to get a nominal resolution of 0 01 but because of sampling error this is only an approximation of the true P value The number of background points you use should be the number of points you have for species B if you have 25 points for A and 50 for B you compare the actual occurrences for A to 50 randomly chosen background points from the area of species B Once these settings have been made they are added to the queue by pressing add this analysis You would then want to make the same comparison in the opposite direction compare the points for B to the background for A When all analyses have been added press the Go button The output from these analyses looks like that of the identity test and is analyzed similarly with one exception a pair of species may be either more or less similar to each other than expected based on the suite of habitats available to them so this is most often treated as a two tailed test One conceptual issue that arises with this test is the definition and justification of what we consider background In some cases e g making comparisons between island endemics or between other allopatric taxa separated by geographical barriers the appropriate geogra
27. very confusing if you don t keep them organized You can create a directory in the same fashion that you would create any new directory with your operating system Select this directory in ENMTools using the Output directory button 3 necessary only if you have selected the Species with data option Select a set of layers for projection If you are using the Species with data SWD option you will need to have a set of layers to project onto If you are using SWD be sure to set your Projection directory button 4 Specify location of Maxent Many of the analyses conducted in ENMTools are done in association with the program Maxent For this collaboration to be successful ENMTools needs to know where Maxent is You can set this location using the Maxent jar file button Make sure you select Maxent s jar file not the bat file Due to changes in the command line arguments for later versions of Maxent ENMTools also needs to know which version of Maxent you re using Select the appropriate radio button at the bottom for your Maxent version 5 Select the suitability measure that you want to use for all Maxent analyses Maxent is capable of providing output in the form of several different types of suitability measures To specify the option used for analyses conducted in Maxent select your preferred option using the suitability measure radio button 6 Set Maxent visibility option You have the option of
28. way Warren and Phillips in prep 23 The retain X jackknife is primarily useful when comparing measurements on pseudoreplicates between species that have different sample sizes e g making comparisons of niche breadth with jackknife estimates of uncertainty One might generally expect that estimates of niche breadth would demonstrate some correlation with sample size particularly when sample sizes are small For this reason it makes more sense to compare species using identically sized subsets of their occurrence points 24 ll d ii Spatial cross validation ENM measurements Hypothesis testing Resampling Options Split species along a straight line for cross validation Files containing occurrences Add files Import file list Save file list Clear file list Number of replicates Proportion of records to retain for testing Run Maxent GO These two analyses use the range breaking methods described above section IlIl c to randomly partition the data for a single species into two non overlapping data sets The program then uses one of the partitions as training data and the other as test data for a Maxent run These analyses can be useful in studies of sample selection bias and may be relevant to studies of model transferability in some situations but see Phillips 2008 Users simply specify a list of files to analyze the proportion of records to withhold as test data and the number of replica
Download Pdf Manuals
Related Search