C. Cam Muir, R.F. Watkins, and T.T. Ha, Map Simulations for Testing a Palaeo-Migratory Hypothesis

------

Map Simulations, Based on Bathometric Data and Information about Glacially Induced Sea-Level Changes, Are Used to Test a Palaeo-Migratory Hypothesis Inferred from Orang-utan Population Genetics Data

C. Cam Muir, R.F. Watkins, and T.T. Ha
Simon Fraser University

------

Introduction
The orang-utan (Pongo pygmaeus) was once distributed throughout south Asia. The modern orang-utan is believed to descend from the common ancestor of all great apes (including Homo) through a lineage including Dryopithecus, Sivapithecus, and Ramapithecus (Andrews 1982; Cronin 1982; Pilbeam 1980; Smith 1980; Temerin 1980). Contemporary populations of the orang-utan are limited to Borneo and northern Sumatra. This change in distribution and the degree of relatedness between subpopulations, or demes [1], are the subject of our study. Availability of migratory pathways between populations plays an important role (in addition to ecological structure, behaviour, climate, etc.) in overall allele [2] or haplotype [3] distribution and population structure. De Boore (1982) suggested, on the basis of morphological characteristics, that there is a significant orang-utan population structure within Sumatra and between Borneo and Sumatra (see Figure 1).

Figure 1. Map of modern South-East Asia

Pairwise differences [4] between haplotypes found among orang-utans indicate some degree of population structure (Muir et al. 1995, also Muir, in review and Muir, submitted; Xhi et al. 1996; Xu and Arnason 1996). The level of divergence noted between some haplotypes indicates significant periods of genetic isolation. There are at least three distinct types found within extant orang-utan populations, which are 5.2-7.8 percent divergent (Muir et al., in review). All three types are found within the Sumatran range and one of these types is found in Borneo. Given significant (up to 2.8 percent) divergence of types found within the Bornean range, sequence identity shared between Borneo and Sumatra is consistent with recent gene flow between the islands. The proposed elevated rate of mitochondrial sequence [5] evolution (Muir et al. 1995, and in review) further supports a recent migration model since gene flow is evident within the time frame in which mutations arise. In either case, persistent sequence identity is unlikely to follow protracted genetic isolation. Although less dramatically illustrated, the same can be said for the populations within Borneo. There is a variety of mitochondrial and nuclear [6] DNA loci for which individuals are more similar to individuals in seperate demes than they are to deme mates (Muir, submitted). A deme is loosely defined as a group of individuals within which interbreeding, and therefore gene flow, is confined.

Two subspecies of orang-utans have long been recognised. Recently, Xu and Arnason (1996) and Zhi et al. (1996) have suggested these subspecies be elevated to separate species status. Muir et al. (1998) argued, among other things, that molecular data were inadequate for defining a species. Furthermore, data from 13 mitochondrial and nuclear genetic loci, taken from a much larger sample set, tell a very different story than has been presented previously (Muir, submitted; Muir, et al. 1995, and in review). There is a variety of mitochondrial and nuclear loci by which some Sumatran orang-utans are more similar to Bornean orang-utans than they are to other Sumatrans. There are even Sumatran orang-utans that share exact sequence identity with Bornean orang-utans (Muir, submitted; Muir, et al. in review). The discovery of individuals on different islands sharing sequence identity, while being different from others on the same island was an unexpected result that begged explanation. Invoking the use of models of changing sea level to explain the emergence of land bridges both chronically over the past million years, and as recently as 12 kya (thousand years ago) provides a hypothetical route across which migration could have occurred.

The availability of migratory pathways is determined by a number of ecological, geophysical, and oceanographic factors. Since these factors have undergone dramatic changes through geological time, an understanding of palaeogeography, and palaeoecology may aid understanding contemporary orang-utan population structure. Bathometric data sets may be used to create 3-D topology map simulations. Inaccuracies in historic representation of bathometric data which may result from erosion/deposition, tectonic dynamics including subduction and volcanism, are, in part, 'hidden' within bathometric data and predictions of relative prehistoric sea level (J. Clague, pers. comm.). By creating a series of topographic maps on which sea level changes are taken into account, a videotape clip can be assembled which shows changes in land exposure above sea level. We created such a videotape of the area including Borneo, Java, Sumatra, and the south Asian mainland for a time period including the Wisconsin glaciation (~60 kya to present). Putative migration rates, implied by population genetics data, were tested with the videotape simulation.

Methods
Digital land height and bathometric data for the region including Borneo, Java, Sumatra, and the mainland of south Asia including Peninsular Malaysia and Vietnam were downloaded from the US National Geophysical Data Center web site <http://www.ngdc.noaa.gov/mgg/global/seltopo.html>. A 709 by 421-pixel data set was selected which included a region from latitude 27 degrees north to latitude 8 degrees south, and longitude 127 degrees east to longitude 68 degrees east at a resolution of five arc-minutes per data point. The data set consisted of a 1,480,003-byte [7] space-delimited text file of elevations relative to sea level ranging from -9394 to +4115 metres. The data were then reformatted to a Digital Elevation Map (DEM) for the computer program Vistapro® version 4 (RomTech Inc.<http://www.romt.com/index.html>). Vistapro®'s DEM format is proprietary and not directly compatible with the DEM standard of the US Geological Survey (USGS). Each elementary manipulation of the data set was accomplished with a Microsoft DOS QBASIC® program written for the purpose.

Since Vistapro® only accepts land heights in a range of integer values from 0 to 32,767, the values in the file were first scaled by adding 9394 to each value in the matrix resulting in values ranging from 0 to 13,509 metres at one-metre resolution (Appendix 1A). Resolution can be increased to any desired value by simply scaling the desired range of land heights and setting the values below to zero and above to the maximum. An increase in resolution was, in our case, deemed unwarranted given uncertainties in estimates of historic sea levels, various unknowns with respect to details of plate tectonic dynamics, and geographic resolution of the map itself. In other words, we did not want to pretend that we had more resolving power than our data actually provided. Following these adjustments, the present-day sea level was set to 9,394 metres and all alterations of sea level in Vistapro® were made relative to this value.

Each row in the file was then reversed, front-to-back, in order to provide correct orientation for data in the Vistapro® file format (Appendix 1B). Conversion to Vistapro® DEM format was performed with the public-domain program Z2DEM which accepts a maximum 258 row by 258 column space-delimited text file of land height values and converts them to a Vistapro® DEM file. In order to accommodate the file size limitation the original file was split into four smaller files (we discarded the eastern most portion of the file), two of 258×258 points and two of 258×163 points (Appendix 1C). Each file was then converted to a Vistapro® DEM file using Z2DEM.

The four Vistapro® DEM files were then reassembled as a contiguous landscape in Vistapro®, setting the Creation/Set Landscape Size option to 512×512, importing the files and manually placing each map element using File/Open Landscape/Manual Placement. The resulting landscape was saved for further use. Sea level changes were made relative to the 9394 value using Water/Sea level setting. The landscape heights were then scaled down by ten times by setting the Manipulation/Vertical Scale to 0.1. This was done in order to give a visually acceptable approximation to the land heights, though a considerably smaller scale value would have been necessary in order to give true land height measures relative to the scale of the map.

In order to generate the animation a script file was generated with another QBASIC® program which was used to specify either an AVI-format animation file for computer use, or a series of TGA-format images for transfer to videotape (Appendix 1D). All parameters remained the same from frame to frame with the exception of sea level, which was altered as a linear interpolation of the sea level at the last glacial maximum (LGM) (125 metres lower than at present). No attempt was made to make elapsed videotape time proportional to actual elapsed time.

Transfer of AVI-format animation file to videotape requires a NTSC scan converter (Averkey3®; see <http://www.aver.com>) and a videotape recorder (VTR). The quality is not optimal, but this method is cost effective. To achieve a CR601 broadcast quality, a digital video-disk recorder, movie-editing software, and a Betacam SP format VTR are required. Perception® video recorder (PVR) <http://www.dps.com>, Adobe Premier® editing software <http://www.adobe.com>, and a Sony BVW- 50® VTR <http://www.sony.com> were used in this case. The hardware and software are available in both IBM PC and Apple Macintosh® platforms. A series of TGA files must be rendered at 720×480 pixel resolution. These files are then imported into the PVR to be converted into one single AVI-format file using PVR compression CODEC. These AVI files are edited with Premier® to complete one movie. The movie must be played back via PVR player software to achieve the smoothness of the animation which can then be recorded into the VTR. This master tape can be used for mass transfer to VHS format.

(All of the steps required to proceed from digital elevation data to finished video are presented in the flow chart in Appendix 2.)

Observation and discussion
In running the changing sea levels video (see AVI 1; appropriate browser plug-in or external viewer required), it is immediately obvious that shallow channels between the islands of Sumatra and Java, Sumatra and the Malay Peninsula, and Sumatra and Borneo, provide the first and longest-sustained land bridges. This observation is consistent with observed sharing of mitochondrial haplotypes (unpublished data) in south-western Borneo and Sumatra. As the sea level continues to drop (to -125 metres at LGM), the land bridge continues to expand to comprise finally all four land masses from south-eastern Borneo and Java, to south-eastern Vietnam and north-western Borneo as a single land mass.

During much of this time this land mass was covered by tropical woodland or tropical rainforest (Vander Kaars 1990; see also <http://www.esd.ornl.gov/ern/qen/new_eurasia.html>), providing habitats similar to contemporary habitats. The interruption of this ecotype by the rising sea level, a consequence of receding glaciers, was maintained even during a period of isostatic rebound. During a period of several hundred years, the continental plates, buoyed by the removal of glacial weight, rose to an elevation which caused the re-emergence of the land bridge until accumulated melt water from the glacier again covered this bridge. During this period, the ecotype for the intervening land mass was open savannah (<http://www.esd.ornl.gov/ern/qen>). Although the possibility of migration remains, it is unlikely that significant migration of the orang-utan (an arboreal animal usually confined to diptocarp tree canopy) occurred over hundreds of miles of savannah.

Open ocean is not the only geographic barrier to migration of orang-utans (or other terrestrial organisms) within and between island populations. Currently, and by extension historically, migrants are probably limited by significant rivers and mountain ranges. Insight into the presence of such land forms, not visible in the sea levels videotape, can be achieved by constructing a 'fly-by' video. This is achieved by lowering the altitude of the 'point of view' or camera position to approximately mountain altitude and creating a video-file (AVI) in which the camera position is trained along a circumference of the area of interest (see AVI 2; appropriate browser plug-in or external viewer required). These manipulations are made through the PATH menu. In viewing the video-file, land forms are recognisable and may help to predict blockage of potential pathways for migration, e.g. major river in south-eastern Borneo, east of Tanjung Puting (across from the southern tip of Java).

Migrations proposed to explain contemporary population structure are predicted to have occurred over more than a single glacial epoch (Muir, et al., in review). This prediction is based on the shared identity of a Bornean haplotype and a Sumatran haplotype, in addition to variation in pairwise differences between Bornean and Sumatran haplotypes.

We suggest that construction of three-dimensional maps in concordance with available bathometric, tectonic, and palaeoecological data will allow a more complete understanding of historical influences on contemporary population structure to be attained.

Acknowledgements
We would like to thank J. Fletcher, Jane and Gordon Tribble, Hugues Faure, Sander Van der Kaars, Peter Kershaw, Lloyd Burckle, Wyss Yim, Birute M.F. Galdikas, OFI, and Researchers in Quaternary Science e-mail list <quaternary@morgan.ucs.mun.ca>. Research was supported in part by a NSERC research grant to A.T. Beckenbach.

------

Appendix 1

1A: QBASIC® Programme to scale digital elevation data to positive values

	MAPMIN = 9394
OPEN "asiase.txt" FOR INPUT AS #1
	OPEN "output.txt" FOR OUTPUT AS #2
		WHILE NOT EOF(1)
			A$=INPUT$(1,1)
			IF NOT (A$=" ") THEN B$ = B$ + A$
			IF A$ = " " THEN
				B = VAL(B$)
				B = B + MAPMIN
				B$ = STR$(B)
				PRINT#2, B$; "," ;
				B$ = "" : A$ = ""
			END IF
		WEND
	CLOSE #2
	CLOSE #1
	PRINT "Done..."

1B: QBASIC® program to reverse line order of digital elevation data

	DIM dat$(800)
	OPEN "input.txt" FOR INPUT AS #1
	OPEN "output.txt" FOR OUTPUT AS #2
	FOR J = 1 TO 421
		FOR I = 1 TO 709 STEP 1
			Skip:	INPUT#1, dat$(I)
			IF dat$(I) = "" THEN GOTO Skip
		NEXT I
		FOR x = 709 TO 1 STEP -1
			IF dat$(x) = "" THEN STOP
			PRINT#2, dat$(x);",";
		NEXT X
	NEXT J
	PRINT#2, ""
	PRINT "Done..."
	CLOSE #1
	CLOSE #2

1C: QBASIC® program to split digital elevation data into subfiles for conversion

	OPEN "output.txt" FOR INPUT AS #1
	OPEN "map1" FOR OUTPUT AS #2
	OPEN "map2 FOR OUTPUT AS #3
OPEN "map3 FOR OUTPUT AS #4
OPEN "map4" FOR OUTPUT AS #5
OPEN "map5" FOR OUTPUT AS #6
OPEN "map6" FOR OUTPUT AS #7
FOR y = 1 TO 258
	FOR x = 1 to 258
		INPUT#1, a$
		PRINT#2, a$;" ";
	NEXT x
	FOR x = 1 to 258
		INPUT#1, a$
		PRINT#3, a$;" ";
	NEXT X
	FOR x = 1 to 193
		INPUT#1, a$
		PRINT#4, a$;" ";
	NEXT X
	FOR z = 194 TO 259
		PRINT#4, "0 ";
	NEXT z
NEXT y
FOR y = 1 TO 162
	FOR x = 1 to 258
		INPUT#1, a$
		PRINT#5, a$; " ";
	NEXT x
	FOR x = 1 to 258
		INPUT#1, a$
		PRINT#6, a$; " ";
	NEXT x
	FOR x = 1 to 193
		INPUT#1, a$
		PRINT#7, a$; " ";
	NEXT x
	FOR z = 194 TO 259
		PRINT#7, "0 ";
	NEXT z
NEXT y
PRINT "Done..."

1D: QBASIC® program to generate Vistapro® script file for animation

REM This program generates a Vistapro Script File which can be used to
REM generate the animation.  Note that the landscape is reloaded each
REM time through the loop because changing the sea level deletes all the
REM land levels below sea level.  This isn't as necessary when the sea
REM level increases, but it makes for more readable code. 9394 is sea level
REM for the scaled data set.  The zvals are increments of sea level
REM change (ie, we're going to go down 125 meters over the next 60
REM frames - I do the math to make it easy to read).  The slight differences
REM from starting Z values in different sections are not significant at the
REM scale of this map.  I felt that a more carefull approach to setting the
REM sea levels wasn't justified given the high inherent errors of the sea
REM level estimates themselves.  The script file generated produces
REM 720x480 files - appropriate for NTSC video.  In Vistapro the script file
REM is called from the Animation tab - click on "Create TGA Series".

OPEN "vid1.scr" FOR OUTPUT AS #1
PRINT #1, "Vista Script File"
PRINT #1, "DefaultDirDEM d:\graphics\vp4"
PRINT #1, "DefaultDirPCX d:\graphics\vp4"
z = 9394
zval = 125 / 60
        FOR i = 0 TO 59
                PRINT #1, "LoadLandscape borneo.dem"
                PRINT #1, "SeaLevel ";
                a = ((INT(10 * z) / 10))
                a$ = STR$(a)
                PRINT #1, a$;
                PRINT #1, " 1"
                PRINT #1, "Tree1None"
                PRINT #1, "Tree2None"
                PRINT #1, "Tree3None"
                PRINT #1, "Tree4None"
                PRINT #1, "CameraXYZ 120 15300 7100"
                PRINT #1, "TargetXYZ 6360 8760 0 "
                PRINT #1, "SkyOff"
                PRINT #1, "HorizonOff"
                PRINT #1, "VerticalScale 0.1"
                PRINT #1, "ImageSize 720 480"
                PRINT #1, "Render"
                PRINT #1, " "
                z = z - zval
        NEXT i
z = 9394 - 125
zval = 60 / 34
        FOR i = 1 TO 34
                PRINT #1, "LoadLandscape borneo.dem"
                PRINT #1, "SeaLevel ";
                a = ((INT(10 * z) / 10))
                a$ = STR$(a)
                PRINT #1, a$;
                PRINT #1, " 1"
                PRINT #1, "Tree1None"
                PRINT #1, "Tree2None"
                PRINT #1, "Tree3None"
                PRINT #1, "Tree4None"
                PRINT #1, "CameraXYZ 120 15300 7100"
                PRINT #1, "TargetXYZ 6360 8760 0 "
                PRINT #1, "SkyOff"
                PRINT #1, "HorizonOff"
                PRINT #1, "VerticalScale 0.1"
                PRINT #1, "ImageSize 720 480"
                PRINT #1, "Render"
                PRINT #1, " "
                z = z + zval
        NEXT i
z = 9394 - 60
zval = 40 / 56
        FOR i = 1 TO 56
                PRINT #1, "LoadLandscape borneo.dem"
                PRINT #1, "SeaLevel ";
                a = ((INT(10 * z) / 10))
                a$ = STR$(a)
                PRINT #1, a$;
                PRINT #1, " 1"
                PRINT #1, "Tree1None"
                PRINT #1, "Tree2None"
                PRINT #1, "Tree3None"
                PRINT #1, "Tree4None"
                PRINT #1, "CameraXYZ 120 15300 7100"
                PRINT #1, "TargetXYZ 6360 8760 0 "
                PRINT #1, "SkyOff"
                PRINT #1, "HorizonOff"
                PRINT #1, "VerticalScale 0.1"
                PRINT #1, "ImageSize 720 480"
                PRINT #1, "Render"
                PRINT #1, " "
                z = z + zval
        NEXT i

CLOSE #1

------

Appendix 2: Flow Chart Detailing the Process of Video Creation


------

References

Copyright © C. Cam Muir, R.F. Watkins, and T.T. Ha 1998

------

contents masthead e. mail issues index contributions

editorial research articles features forum field notes reviews state of... wise words games info links

------

Copyright © assemblage 1998