Lacey Jane Wolfe
Russell Bither-Terry
Lab 5
Geography 491
1) What relationships exist between farm characteristics (in the attribute table) and the percent guava by farm?
Percent Guava by Field vs. Most Recent Year Cleared
As you can see in Figure 1, we found a correlation between the percent guava per field and the year that the farm was most recently cleared. Fields that hadn’t been cleared for ten or fifteen years showed the highest percentage of guava, and fields that had been cleared in the past five years mostly showed the least percentage guava.
Figure 1: Percent Guava and Year Most Recently Cleared by Field
Most invasive plant species are opportunistic, which means that they grow quickly after an area has been cleared. They can “shade out” more slowly-growing species, or block their access to sunlight. If this were true for guava, we would see an immediate response to the clearing of fields (high percent guava in the fields cleared in the past five years), and then guava would continue to be a strong presence even as other, slower plants have a chance to grow (in the “most recently cleared” years 1985-2000).This analysis suggests that guava is able to compete with its community on some other basis.
Percent Guava by Field vs. Size of Field
In Figure 2, we see that as field size (area) increases, the tendency to have greater guava invasive presence also increases. Most of the small fields have less than forty percent guava presence, especially those fields under 500,000 square meters. There are fields of this size with up to ninety percent guava presence and some variation between. In fields of greater than 500,000 square meters, all fields experienced invasive guava at more than thirty percent.
Figure 2: Percent Guava and Size of Farm in Square Meters
Percent Guava vs. Percent Pasture
We also examined the correlation between percent guava and percent pasture (see Figure 3).
Figure 3: Percent Guava vs. Percent Pasture
Generally, we found that as percent guava increases, percent pasture decreases, and vice versa. The guava appears to be pushing out not only the farmers, but also grazing animals. Perhaps the guava creates a thicket that is impenetrable for the animals and so it is crowding them out.
Percent Guava vs. Land Use
Finally, we compared the average percent guava with land use categories. We found that for all areas observed, the overall average percent guava was 44%. When you break down the areas by land use, we find a different story. In land designated as fishing, guava was recorded on average to be 65%. In land designated for tourism, guava was recorded at 50%. For land marked as neither fishing nor tourism, the average guava spread was only 30%. We believe that fishing areas had the highest guava presence because guava likes marshy areas. The land use demarcations are varied enough to indicate that this is not coastal fishing but fishing on rivers and lakes. Tourism is the land use with the next-highest guava presence. This is probably because activities like hiking spreads the guava seeds around, creating a positive feedback loop between tourism and guava presence. The areas that are marked as neither tourism nor fishing are most likely being used only for agricultural purposes. As farmers are pushed out of pastures that have been overtaken by guava, they seek out the few areas left that are not invaded.
2) What, if any, proximity relationships exist between percent guava by field and (a) roads and (b) buildings?
(a) Percent Guava by Field vs. Proximity to Roads
For our analysis of roads and buildings please refer to Map 1.
What is apparent about the roads is that there appears to be more guava near them. In some places there is even a strip of guava alongside a road, with cleared patches further away from the road. We made the roads thicker so they will show up over the farm field boundaries, and this pattern was even more apparent when we looked at thin road lines without those boundaries. The major exceptions lie outside of the agricultural area (visually one sees the road passing through a white patch. The other major exception is farm field number 29, where the areas near the road have less guava and the areas further away have more. I suspect that this is because the roads make it easier for guava seeds to be transported.
(b) Percent Guava by Field vs. Proximity to Buildings
For our analysis of roads and buildings please refer to Map 1.
The overall pattern with the buildings is that there is less guava next to most buildings. Perhaps it is being cleared manually or there is more cultivation of crops and/or grazing of livestock taking place near the buildings.
3. Assign each farm a category of “abandoned,” “partially abandoned,” or “active,” based on percent guava present and the farm characteristics in the attribute table. Are fields or farms next to abandoned fields/farms more likely to contain a higher percentage of guava?
To classify the usage level of farms we first added the percent used for crops to the percent used for cattle. We then classified the farms as abandoned, partially abandoned, and active. We decided that anything over fifty percent used was active, anything above 10 percent was partially active and the three farms that were 0 to 10 percent used were abandoned. Of course this is subjective, but we suspect that most people would draw similar cut points.
4. Is there spatial clustering of guava in the agricultural zone or in individual fields, and does this spatial clustering provide clues to land practices that could affect guava distribution?
The pattern that emerges in Map 2, which shows the guava and the farm boundaries, with each farm labeled by usage level is that, while not all active farms have low(er) guava coverage, all the farms with lower guava coverage are active. This suggests that the active use of the land for cultivation and grazing cuts down on the presence of guava.
5. Is there a relationship between guava in the park and distance to agricultural zone?
As you can see in Map 3, presence of guava decreases the further you get from the park zone (white space in the lower left-hand corner). The distance increases with the class number, so that the magenta ring (value equals 2) closest to the park zone has far more pixels than in dark blue ring (value equals six).
6. Is there a relationship between guava distribution in the park and percent guava in fields bordering the park?
After assigning each pixel within a 500 meter range outside the agricultural area to the farm closest to it, we joined that layer to the farm_fields layer in order to analyze the characteristics of which farm fields had more or fewer guava pixels associated with them in the park. In other words, each pixel is assumed to be most influenced by the farm field closest to it. A good number of farm fields has values of zero. This means they weren’t closest to any guava pixels in the park. We are interested in farm fields on the edge of the agricultural area that impact the park, and thus they were eliminated. Farm field number 1 had an extremely large value, more than ten times the next largest value. Looking at the map at this stage in our analysis (we did not export this map) it did not appear to be the case that it actually had that many more pixels. Whether it is an extreme data point or due to faulty data we thought it best to exclude it as well.
The first thing we consider is if areas of the park closest to more recently cleared fields have fewer guava pixels. There is some evidence for this, as shown by figure four.
Figure 4: Guava Count Vs. Year Most Recently Cleared
The next variable we look at is the area of farm fields. Here is the graph for that:
Figure 5: Guava Count vs. Farm Field Area
There are two important things worth noting regarding this graph and the apparently impressive R-squared value. The first is that it is probably affected by the two obvious outliers. The second is that we should consider the substance of the relationship. Upon reflection one realizes that of course larger farm fields on the border of the agricultural area will have more guava nearest them, they have more area in the park overall—that is both with and without guava—associated with them. That is, even if guava were distributed totally evenly throughout the park we would still see this relationship to some degree simply because bigger fields are the nearest field to more area in the park.
Next we consider the relationship between percentage guava in a farm field and the guava pixel count (which, again, is measuring guava in the park that is nearest the field). Here there is a strong theoretical reason to expect a relationship: if the field has more guava that is more guava to spread into the park. However, this is not borne out by the data. We were for some reason unable to get Excel to make an equation or R-squared for the trend line in this graph, but given how flat the line is this hardly matters (see Figure 6).
Figure 6: Guava Count vs. Percent Guava for Farm Fields
7. How about between guava distribution in the park and characteristics of farms bordering the park?
Doing the same analysis for farms (and again excluding farms that had values of zero for the guava pixel count), we again look at the relationship between percent guava and the guava pixel count. This time we get a fairly strong relationship (Figure 7), which makes sense given the theoretical reason for it (outlined above). Perhaps we did something wrong when we looked at it at the farm field level.
Figure 7: Guava Pixel Count vs. Percent Guava for Farms
Farms where more of the land was used showed lower levels of guava associated with them in the areas of the park nearest to them (see Figure 8). This makes sense, as if the land is used then the guava will be either eaten by livestock or manually cleared during weeding crops.
Figure 8: Guava Pixel Count vs. Percent of Land Used (for Farms)
We can break land usage down into crops and pasture to see if this tells us anything about which is more important. As we see both demonstrate the same trend. The scatter plot for crops (Figure 9) has a slightly better R-squared value than for land use and the model for pasture has a much lower level. So perhaps crops have a stronger impact on clearing guava than using land for pasture. We also look at the number of cattle and the impact that has on guava in the park (Figure 10).
Figure 9: Guava Pixel Count vs. Percentage Crops (for Farms)
Figure 10: Guava Pixel Count vs. Percent pasture (for Farms)
The other data we have about the use of farmland which might help clear guava is the number of cattle on a farm (Figure 11). This has a somewhat better R-squared value, but still not as good a value as that for crops. This suggests that it matters not just how much land is used for pasture, but how intensely it is used—more cattle eat more guava and less of this spreads to the park.
Figure 11: Guava Pixel Count vs. Number of Cattle (for Farms)
Guava Invasion: Preliminary Findings
Invasion in Agricultural Zones
We have observed in our analysis that guava appears to be a slower-growing invasive, due to the relationship between clearing fields and guava presence. Most invasive species grow quickly after an area has been cleared. They spring up fast and then shade out other species. We have found that guava is most present in farms that have not been cleared for ten to fifteen years.
According to our findings, guava will slowly invade a field cleared for farming over a period of ten to fifteen years. This is discouraging because it suggests that fields will be overgrown with the invasive species and will not be useful to farmers for very long unless they re-clear the field. Our analysis of pastures (which we interpreted to be land used to raise livestock) yielded a negative correlation between percent guava and percent pasture, presumably because the guava was crowding out the livestock.
A land use analysis showed that the highest presence of guava was found in areas designated for fishing. We suggest above that this may be because the guava plant likes to be near water, and it thrives there. The presence of guava in area designated for tourism, however, we believe is due to the impact of tourism on the environment. Having people walking through particular areas, leaving waste behind and tracking seeds around, appears to have a positive impact on guava growth. This is a contrast to the agricultural areas, which only experienced a 30% guava presence.
As far as roads and buildings are concerned, the presence of guava increases with proximity to roads but decreases with proximity to buildings. We assume that the correlation with buildings is because the area around buildings is routinely cleared or that it is paved so that the guava cannot take root. The proximity to roads is probably impacting guava growth in the same way that tourism is spreading around seeds and waste with the increased traffic of humans.
On the other hand, human contact with land in the form of crops and livestock appear to help keep land cleared of guava. This makes sense as weeding crops and livestock eating crops would help control
Therefore, in terms of guava invasion in farms, we suggest that proximity to roads and tourism areas are the major sources of invasion. Grazing animals seems to accelerate the invasion. We suggest that establishing buffer zones for farms may be an effective way to delay the invasion of guava. In the past, farmers have used this method effectively in the cultivation of other crops such as corn. For example, farmers will plant a buffer of “hog corn” (corn cultivated for the feeding of livestock rather than to be sold as food for humans) near roads. This prevents their more valuable crops from being damaged by the invasion of weeds.