1   Abstract 

     The bonobos of the Democratic Republic of the Congo (DRC) share much in common with humans and offer much research potential in the study of human evolution, but they are currently endangered and rapidly declining in numbers. We studied a few areas of previously documented, dense bonobo populations in order to determine which type of environment they have the most potential to thrive in--focusing especially on vegetation. We also observed threats to bonobos such as vegetation loss, fire, and urban expansion. Using the remote sensing software ENVI, we created images which explore the conditions in which bonobos live. Our work found that bonobos fare best in areas which have a varied combination of savannah, dense green forest, and rainforest, particularly one area which is not currently protected. We unveiled that there was a reduction of vegetation in bonobo habitats between 2002 and 2007. If change can be seen in five years, bonobos are clearly in danger of losing their land in the near future. Our hope is that further research can apply our findings to broader bonobo population areas and that DRC zoologists can take steps on the ground to determine the cause of and fight back against vegetation loss.

               2   Introduction    

     Bonobos, Pan paniscus, along with common chimpanzees are the closest living relatives of humanity, sharing over 98 percent of genetic material (Franz De Waal, 1997).  Though bonobos and chimpanzees are equally related to Homo sapiens, their separate, distinct physical and behavioral characteristics give contrasting clues as to what the original ancestors of humans may have been like.  Where in chimpanzee societies, males dominate; in bonobo societies, mothers have authority.  Where chimpanzees commonly display aggressive behavior, bonobos frequently display sexual, affectionate behavior.  However, due to the validation of bonobos as an official species (unique from chimpanzees) long after the studies performed and papers written with chimpanzees as the sole closest human relative, humanity is tainted with the violent, tool-using, male-centered image of chimpanzees.  The peaceful, egalitarian lifestyle of bonobos was over-looked and, consequently, not given a chance to balance out the established connection between humans and chimpanzees.  There is still much that can be learned in regards to human evolution and the relationships between humans and bonobos, but there is a problem: bonobos are in decline.  The International Union for Conservation of Nature and Natural Resources (IUCN) currently lists Bonobos as endangered, citing commercial poaching as the most potent contributor to the species’ downward population trend.  Other great threats are those related to expanding human activities, such as agriculture, logging, mining, and residential development.

Figure 1

     One main reason as to why bonobos were first overshadowed by chimpanzees in the eyes of science is that bonobos are difficult to locate.  The only region in which bonobos are known to inhabit is in the central Congo Basin of Equatorial Africa in the Democratic Republic of the Congo.  To reach the bonobo habitats of humid primary, secondary, and seasonally inundated swamp forests was, and largely still is, a strenuous journey by foot or primitive watercraft.  Given this remoteness, current figures for total bonobo population numbers range from 29,500 to 50,000 -- with underestimation still being contended (IUCN, 2010).  Hand-in-hand with impenetrable forest, the DRC is a country plagued with warfare and economic instability.  Though, the DRC has its own official conservation administration, Congolese Wildlife Authority (ICCN), much of the country’s creation and management of natural parks and reserve areas is supplemented by non-governmental organizations and research projects.  Much of the focus of this project will center around Lake Tumba and Lake Maindombe of northwestern DRC, which are part of the Congo River Basin.

     Research and conservation projects in this area are largely driven by the World Wildlife Fund for Nature (WWF), where the first step is to identify existing populations, inhabitable land, threats, and opportunities for protective action.  The research paper, “The Bonobos of the Lake Tumba – Lake Maindombe Hinterland: Threats and Opportunities for Population Conservation ,” (2008)  details the results of the task and reveals valuable findings that expand the range of documented bonobo habitat and presents the idea that particular landscape combinations may allow for higher community densities.  A similar study in Lukuru (southeast of Lakes Tumba and Maindombe) by J. A. Myers Thompson conducted in 2001 verifies that bonobos have, indeed, seasonally occupied drier savannah habitats in addition to forests.  Building off of the above 2008 report, "Bonobo Food Items, Food Availability and Bonobo Distribution in the Lake Tumba Swampy Forests, Democratic Republic of Congo" (2009) is a comprehensive comparison of permanent bonobo study sites regarding population densities and food availability.  The work concludes that high populations may be supported by both high fruit and high terrestrial herbacious vegetation (THV) availability provided for by mixed habitats -- forest-savannah mosaics.  The IUCN has published a table comparing these bonobo densities HERE.  Building off of this concrete information, we wish to utilize remote sensing images to compare the vegetative commonalities and differences between documented bonobo populations, thus confirming the report's findings.  In doing this, we hope to visualize the ideal bonobo habitat that may allow for high population densities, thus creating a model for further bonobo population research.  The use of remote sensing software has potential to make this research more efficient by pin-pointing likely areas first, rather than physically scouting.  We wish to provide a way of identifying where the needle is in the haystack.  To protect all bonobos, it is crucial to know where they all are.

Fig. 2  and Table 1 taken from "Great apes in the Lake Tumba landscape, Democratic Republic ofCongo: newly described populations."  Note that the highest Bonobo density is found in the 5b habitat block.
 

              3   Methods

 We based our study areas on the "Great apes in the Lake Tumba landscape" article, which divided the Congo's bonobo population into four primary areas--regions 2, 3, 4, and 5b. All four of these regions are located in the western portion of the Democratic Republic of Congo just east of the Congo River. We focused especially on region 5b (located between the villages of Malebo and Nguomi), since it has by far the highest concentration of bonobos.

We obtained images to study NDVI (the amount  and quality of vegetation) from the Maryland Global Land Cover Facility website. Using the Earth Science Data Interface, we downloaded the most recent ETM+ images from Landsat 7 for the 4 locations listed in Table 1 above, in addition to other locations with documented population density estimates. Crucial to NDVI analysis, we downloaded Landsat 7 images that included band 3 (red) with a spectral resolution of .63-.69 microns and a ground resolution of 30m as well as band 4 (infrared) with a spectral resolution of .75-.9 microns and a ground resolution of 30m. Panchromatic images (band 8) were also downloaded, as the more focused spatial resolution can help to identify human settlements (which may pose threats to bonobos). In order to analyze the thermal conditions of each community, we downloaded the Landsat 7 image that included band 6 (thermal) with 10.4-12.5 microns and a ground resolution of 60m. All of the downloaded images were produced by the USGS.

We edited Landsat 7 images exemplifying NDVI and temperature across each location using the ENVI Remote Sensing program. For NDVI, we first opened our images (Bands 3 and 4) in ENVI (choosing RGB and setting the bands to upload as R: band 4 G: band 3 B: band 4 before opening image in display) then subsetted the image to make vegetation samples for each location. For some Landsat images, it was necessary to replace bad values, mitigating the consequences of the line drop problem. We then went to the power bar, chose transform, then clicked on NDVI. After choosing our subset image, we chose Landsat TM and changed the red band value to 2 and the NIR band value to 1. After renaming the file before saving, we then opened the image in a new display. The image displayed was black and white.  We then created a density slice overlay of the image. showing the highest amount of vegetation in dark green and black, and the lowest amount of vegetation represented by red and white (bare areas) .  Yellows represent intermediate areas.  For each sample, we utilized the ENVI tool to create horizontal profiles.  The profiles are graphs of NDVI values (-1.0 - +1.0, the y-axis) from west to east of the sample area. 

To study how the most populated area of bonobos (5b) has changed, we opened NDVI images of the region from 2002 and 2007 in ENVI, using the same steps we used to create NDVI images above. In the main ENVI bar, we selected Tools, then Basic Tools, then Change Detection, and finally Compute Difference Map. Using the NDVI images, we opened a difference map which showed which areas had developed more vegetation and which areas had lost vegetation between 2002 and 2007.

Besides studying vegetation, we also looked at elevation data. We obtained the Landsat 7 ETM+ images from the Maryland Land Cover Data ESDI site. Next, we uploaded the images in ENVI with Band 3 representing the Red band, Band 2 representing the Green band, and Band 1 representing the Blue band in order to make the image visible. Using that image, we went to the Tools bar and selected 3D Surface View. Using Band 2 as the input band, we created an image subset of the 5b region. Under the Options tab in the subset, we selected Surface Controls and changed the Vertical Exaggeration to 1000 to visually show the area's elevation differences. We zoomed in and rotated the image to best visualize the elevation. Finally, we changed the background color to black under the Options tab. To numerically show the elevation, we selected Profiles under the Tools tab and created either Y Profiles or X Profiles for each area. We repeated these steps for the images of regions 5b, 3, 2, and 4.

Finally, we studied night light data using Professor Gillespie's night light image of the world from 2009. We opened the image in ENVI, then created a spatial subset (File -> Save Image As -> Image File -> Spatial Subset). We opened the subset and used the Cursor Location/Value tool (Tools -> Cursor Location/Value) to create a image in Microsoft Paint that showed the location of areas with heavy night light (Mbandaka and Mushie) in relation to our four study areas.

   4   Results

 

Fig. 3: Using Google Earth and Microsoft Paint, we created a crude replication of the 5b, Malebo-Nguomi location to assist with locating the region using satellite images in ENVI.  Below is a Landsat 7 ETM+ image taken in 2002.  The red dots indicate the same "local places" thumb tacked above.  This is how we gathered are NDVI samples for each location.

Fig. 4

4.1    Vegetation

Sample Density Slice parameters:

      -1.00 - -0.75:   Blue
      -0.75 - -0.50:   White
      -0.50 - -0.25:    Red
      -0.25 - 0.00:     Yellow
       0.00 - +0.25:   Yellow 2
      +0.25 - +0.50:  Green
      +0.50 - +0.75:  Green 3
      +0.75 - +1.00:  Black

Location 5b (Landsat ETM+, January 2, 2007)
Density per square km: 1.8 - 3.4 (The most densely populated in DRC!)

Fig. 5: ABOVE: This sample clearly illustrates the forest-savannah mosaic.  In yellow is the savannah -- grassland with trees spaced widely apart.  The presence of darker yellow are areas that are greener than grasslands, but are not quite forests - such as shrublands.  In this case the dark yellow demonstrates the forest repopulating the savannah ("Bonobo Food Items, Food Availability p.21), a unique phenomena that may be providing for increased bonobo density   The green is rainforest.  The speckling of dark green illustrates denser, greener forest.  What is unique is the size and quantity of savannah in relation to the forest.  Having a mix of the two biomes provides year-round food availability (fruits in the savannah and THV in the forest, as well as savannah).  Here we see that the combination is vast, consistent, and even.  The boundaries of variation are relatively solid.  BELOW:  A horizontal profile from the center of the sample above (plotted NDVI values).  This profile cements that values in the darker yellow region (0.00 - +0.25) are in the majority, in comparison to less green savannah values (-0.25 - 0.00).  Dense, green forest are also common (values above +0.50).  Again, the large spaces between forest and grassland are easily observable.

Fig. 6

Fig. 7

Fig. 8: ABOVE: Another image sample from 5b that illustrates the forest-savannah mosaic, where the forest is slowly repopulating grasslands.  The mixture of biomes is still relatively large, consistent, and even with defined boundaries.  Denser, greener forest is also widely present.  BELOW:  The profile keeps in uniform with the first sample, with values rising and falling between -0.10 - +0.60, demonstrating the changes from forest to savannah areas.

Fig. 9

Fig. 10

Location 3 (Landsat ETM+, July 10,2006)
Density per square km: 0.24 - 0.29 (Second most dense in Lake Tumba region)

Fig 11: ABOVE:  For this location there is a significant increase in shrubland areas (dark yellow), verified by observing the region on Google Earth.  The forest-savannah mosaic has lost consistency and defined boundaries (disintegrating appearance).  There is still a noticeable presence of darker green forest.  BELOW:  The profile visually demonstrates the difference between the 5b locations.  There are few dramatic rises and drops between grassland and forest values.  The prevalence of shrubland creates a gradual rise and decline in values, with drops below 0.00 not as common as 5b.

Fig. 12

Fig. 13

Location 4 (Landsat ETM+, July, 16 2005):
Density per square km: 0.24 - 0.29 (Third most dense in Lake Tumba region)

Fig. 14: ABOVE:  In this location we see a significant increase in forested area in combination with shrubland and patches of grassland.  The mixture of savannah with forest resemble location 3 more than 5b, showing that the mixture is less integrated and lacking defined boundaries.  BELOW:  The profile clearly demonstrates the continuity of the area as forest with little gradation into grassland.

Fig. 15

Fig. 16

Location 2 (Landsat ETM+, July, 16 2005):
Density per square mile: 0.25 - 0.29 (Least dense in Lake Tumba region)

Fig 17: ABOVE:  In comparison to the images of 5b, it becomes clear why this is the least dense of the locations.  There is complete absence of dense, greener forest, and is mostly swampy shrubland.  BELOW:  The lack of vegetational variation is clearly illustrated noting values mostly contained between -0.05 - +0.25.

Fig. 18

Fig. 19

Lukuru (Landsat ETM+, August 03, 2005):
Density per square mile: 1.04 (Second most dense in DRC)

Fig. 20: ABOVE:  The above NDVI is of Lukuru, with the second most dense bonobo population estimate in DRC.  Observing the image, we see a return to the forest-savannah mosaic with consistency and defined boundaries.  However, it is obvious that the grassland is not of the same character as the mosaics observed in 5b.  The large patches of red, white, and blue reveal the possibility that the grassland may not be as hospitable to bonobos as the transforming grasslands of 5b.  BELOW:  Again, we see a return to the mosaic pattern of highs and lows, but here the lows are more drastic, commonly reaching below -0.30.

Fig. 21

Fig. 22

4.2   Elevation

Fig 23: This image reflects the elevation in the greater area surrounding region 5b. It unveils the fact that there are other areas east of 5b which have similar elevation to 5b and therefore could also have large populations of bonobos. (These regions were outside the survey blocks of the "Great Apes in the Lake Tumba Landscape" article we based our study areas on.)

Fig 24: ABOVE: Location 5b has a wide range of elevation--a mixture of mountains, hills, and valleys. BELOW: The y-profile indicates that the area constantly switches back and forth drastically in elevation.

Fig. 25

  

    4.3    Change Detection

Fig. 26: ABOVE & BELOW:  The images are change detections for the corresponding NDVI images of Loaction 5b above.  The change detection is facilitated by Landsat ETM+ images from May 12, 2002 to January 2, 2007.  The darker shades of red may indicate where forest has successfully repopulated the grassland, while lighter shades may indicate the process of succession, where the grassland has become greener on the NDVI index.  The areas that are the darkest blue indicate areas where forest has been reduced.  Due to the proximity of villages, this may be human induced.

Fig. 27

4.4   Fire

The link in the heading goes to University of Maryland FireFly website, which shows an updated feed of all fires which are currently burning (within the last five days) using thermal data from the MODIS satellite. As of November 29th, our region of interest has a few small fires--none of which are directly within any our studied bonobo population areas. However, this indicates that it is common for fires to burn in the area and these fires likely sometimes occur directly in bonobo habitat.  The research paper, "The Bonobos of Lake Tumba" (2008) reveals that fire is a tool used by the local cattle industry to maintain the existence of grasslands.  This interaction may actually be preserving the forest-savannah habitat that is providing the highest recorded bonobo density.

4.5    Night Light

Fig. 28: This night light image of our four study areas shows that there is no night light activity in or near any of our four regions. The Congo River (vertical) and Kasai River (horizontal) help to locate these bonobo population areas. The settlements of Mbandaka and Mushie contain the closest night light activity to our regions of interest.

5   Discussion

We were successful in finding a model that represented the vegetative characteristics of location 5b -- a region documented as having the highest density of bonobos per square mile in the Democratic Republic of the Congo.  We revealed this model by examining (comparing and contrasting) the NVDI of our samples from five separate locations.  What was found to be unique about 5b is its arrangement of forest in relation to grassland.  The grassland of 5b is of a higher NDVI index (dark yellow) than other more bare grassland (yellow), as it is in the process of transformation into forest.  This was verified visually on Google Earth, but is also evident as the boundaries between forest and grassland are sharply defined.  In addition, the forest of 5b contain lots of greener forest, represented by dark green.  This may allude to a rich nature of the 5b forest.  All of these aforementioned visual characteristics are validated by the sample's horizontal profiles.  The characteristics of 5b's plotted NDVI profile are unique in that the transitions from 'forest-to-grassland-to-forest' are represented with a relatively sudden drop in values to a relatively sudden increase.  Where the values drop represent savannah (grassland), and where the values are higher is forest.  This fluctuation ranges from -0.1 to +0.60 or higher.  Location 5b is the only location that demonstrates this rising and falling in a repetitive fashion, meaning that there is a lot of forest and a lot of savannah and it changes frequently along a horizontal profile.  In addition, the forest contains consistent patches that are above +0.50.  The other samples did not demonstrate the same characteristics, as they were either devoid of grassland and/or dark green speckled forests, or the variation between forest and grassland was not frequent.  Lukuru, the location with the second highest reported bonobo density nearly matched the profile of 5b, except that the savannah portion is of a bare quality, identifiable by large patches of red and white.

Although we have successfully demonstrated the uniqueness of the environment of Location 5b by creating an NDVI using ENVI, we must admit that we are making a broad assumption that NDVI alone can help to pinpoint where high population densities of bonobos can be found.  In looking at the established research, we see that the locations we imaged have higher yields of fruit and terrestrial herbaceous vegetation in combination than any other permanent study site.  In 5b the high THV availability is accounted for by the precession of grassland into forest.  It has also been noted that some species of the THV available in 5b are high in protein content, thus providing the bonobos with a more balanced meal, allowing populations to flourish.  Our model operates on hope.  We only assume that the NDVI values and pattern of the model represent this availability of year-round fruit and THV availability.  To be able to determine the abundance and health of specific plant species using satellite imagery is beyond the scope of our capabilities.

In addition, there are other factors at play that may combine with food availability to determine high bonobo densities, such as elevation, temperature, rainfall, mobility, and human activities, summed up by the IUCN HERE .  We would have like to investigate these items further, but time was not on our side.  The night light image, while helpful for identifying areas with large amounts of electricity, is not a perfectly effective way of analyzing human impact in a region in which most people live with very limited (if any) electricity. Even though no night light appeared in any of the four bonobo population areas we studied, the villages within and nearby these areas could still be affecting bonobo living conditions.

The model we found can hopefully be utilized to search and find more highly dense populations of bonobos, which may expand the documented bonobo range.  Perhaps bonobos crossed rivers previously thought to be their absolute boundaries?  Or perhaps 5b is the only location where the highest density can be found?  Is it a combination of human and ecological forces that have created this high density?  The world must work with what it already knows to preserve bonobos.  As we recently saw with the Western Black Rhino, extinction is real if the right measures are not taken.  We know what is causing the demise of bonobo populations, and steps must be taken now to begin mitigation, with elimination as the goal.  A strong first step would be to extend the Lac Tumba-Ledima Reserve to include the 5b region.  If not, we call for the immediate creation of a protected area to envelope the entirety of forest-grassland mosaic of the 5b location.

6   References

De Waal, Frans.  Bonobo: The Forgotten Ape.  University of California Press.  May 23, 1997.

Fruth, B., Benishay, J.M., Bila-Isia, I., Coxe, S., Dupain, J., Furuichi, T., Hart, J., Hart, T.,    Hashimoto, C., Hohmann, G., Hurley, M., Ilambu, O., Mulavwa, M., Ndunda, M., Omasombo, V., Reinartz, G., Scherlis, J., Steel, L. & Thompson, J. 2008. Pan paniscus. In: IUCN 2011.IUCN Red List of Threatened Species. Version 2011.2. <www.iucnredlist.org>. Downloaded on 29 November 2011. 

Inogwabini, Bila-Isia and Matungila, Bewal.  "Bonobo Food Items, Food Availability and Bonobo Distribution in the Lake Tumba Swampy Forests, Democratic Republic of Congo."  The Open Conservation Biology Journal.  2009, 3, pp. 14-23.

Inogwabini, Bila-Isia, Matungila, Bewal, Longwango1, M.,  Abokome1, M.,  and Vuvu1, M.  “The Bonobos of the Lake Tumba – Lake Maindombe Hinterland: Threats and Opportunities for Population Conservation.”  The Bonobos: Behavior, Ecology, and Conservation.  Springer.  2008, pp.273-290.

Inogwabini, Bila-Isia, Matungila, Bewal, Longwango1, M.,  Abokome1, M., and Tshimanga wa Tshimanga.  "Great Apes in the Lake Tumba Landscape, Democratic Republic of Congo: Newly Described Populations." Oryx. Oct 2007, Vol 41 No 4, pp. 532-538.

Liping, Di and Kobler, Ben. "NASA Standards For Earth Remote Sensing Data." International Archives of Photogammetry and Remote Sensing. Vol XXXIII, Part B2. Amsterdam 2000, pp. 147-155.