Results and Discussion
Chapter Introduction
Results we produced were either maps, statistical analyses or plots.
By interpreting them, we managed to draw some interesting conclusions.
Additionally, some of the figures produced will be included in the final scientific paper by Professor Macusi.
In this chapter, I'll provide and shortly discuss an example for each of our results.
By the end, you'll leave with a better idea on the spatial distribution of artisanal fishing efforts in the Davao Gulf. Hopefully, these results will end up being useful in planning or reviewing management efforts in the gulf.
Note: In order to protect the information of the research project my work contributed to, not all results will be shared, and discussions won't be given in full.
If you're interested for more information, please read the scientific study by Professor Macusi once it's been published.
Spatial Distribution of Fishing Areas
Spatial Distribution of Fishing Efforts
Maps visualizing the spatial distribution of artisanal fishing efforts were successfully produced, and an example is given in image 7.
Analyzing this map by itself is limited to describing the fishing areas seen within it.
As an example, it can be seen that fishers from Samal Island tend to stay confined within a region no further than ten kilometers from its west coast. The depth of this region is approximately 200 m, suggesting these artisanal fishers primarily focus on neritic species.
In contrast to this, fishers from Malita fish in three approximate areas. The first is a region close to the shore, covering the continental slope. As a result, this regions' depth ranges between 200 to 2000 m. Here, Malitan fishers likely target both neritic and pelagic species. The other two regions are much further from their home port, up to approximately 70 kilometers. One of these is adjacent to the shore of the Don Marcelino municipality (connected to the southern border of Malita) and the other is located near the gulf's connection to the Great Pacific. Presumably, fishers from Malita can travel further than those from Samal Island due to them having a larger region of the gulf available to them and due to them possessing more advanced vessels. Additionally, by assuming fishers choose their fishing regions based upon a trade-off between distance from port and the region's CPUE, we can hypothesize that they travel such great distances because the Catch Per Unit Effort (CPUE, expressed in kg/hrs) in these regions farther from port are significantly higher than that near Malita's coast.
Governor Generoso fishers seem to show a similar distribution of fishing efforts when compared to Malita, suggesting they target similar fish and use similar fishing vessels. In order to keep this section short however, I won't discuss their distribution further.
Overlap Between Artisanal Fishing Areas
By calculating convex hulls for a subset of our GPS data representing only presumed fishing activities, we've also produced maps visualizing the overlap between estimated artisanal fishing areas. An example of such a map is seen in image 8.
In this map, it can be seen that most of the fishing areas near Samal Island are concentrated in a tiny region no further than five kilometers from shore, which corresponds with what was seen in image 7. This was to be expected, as fishers from Samal Island are described to use bagnet fishing gear in addition to a technique which attracts nearby fish. By spreading out their efforts, fishers might dilute the fish when using this technique, thereby reducing their CPUE.
Although fishing areas near Governor Generoso also show an immense degree of overlap at a region approximately 10 kilometers offshore, we don't believe this is for the same reason previously discussed, as these fishers aren't reported to use such techniques. Instead, these fishers were described to primarily use hook and line fishing gear (image 10), and we believe they stick to this region because it naturally
has a relatively high CPUE associated with it.
Another point worth noting is the apparently limited competition between artisanal fishers at regions further from coast. As we saw significant fishing efforts in these same regions in image 7, we believe these offshore efforts are caused by a few artisanal fishers spending a lot of time here as opposed to many artisanal fishers spending a little time in these regions. This might make sense due to commercial fishing vessels also being allowed to operate in the offshore regions (> 15 km from the coast) during the month represented by this map, thereby potentially scaring off most artisanal fishers through competition.
As a result, I won't describe the comparison of maps based upon GPS and interview data here. If you're interested in this however, please read the paper by Professor Macusi once it's been published.
Distance Correlation Analyses
By calculating the distance between the centroids of polygons representing estimated fishing areas (image 8) and the nearest port, we estimated the mean distance vessels travelled.
In our R analyses, we then tested correlations between this estimated distance travelled and variables representing distinct fishers, municipalities and months.
Distance ~ Fisher
The p-value for the Kruskal-Wallis analysis testing for this effect was 7.256e-08, which indicates there's a strong correlation between the distance boats travel and the fisher who owned the boat. The associated eta-squared effect size estimate was 0.395. This value can be interpreted by 39.5% of variance within the distance variable being explained by the variable representing distinct fishers. According to this page describing the used formula in the rstatix package for R, the magnitude of this effect size is high when compared to common interpretations of the eta-squared value in published literature.
For these reasons, we believe the variance in distance travelled between fishers was much larger when compared to the variance seen within a distinct fishers' fishing trips. In other words, we believe this result suggests fishers tended to stay loyalty to their fishing areas, but differed among each other in the areas they were loyal to.
Distance ~ Municipality
A Kruskal-Wallis analysis on this correlation resulted in a p-value less than 2.2e-16, suggesting there's an immensely significant correlation between the distance boats tended to travel and their municipality of origin.
Pairwise comparisons between each municipality were also made. As this pertained doing three tests in total, p-values were corrected with the Bonferroni method. Still, all pairwise comparisons remained significant, with p-values ranging between two levels of significance (image 10).
Notably, these pairwise comparisons also resulted in different estimates of effect size. When comparing the distance travelled between Governor Generoso and Samal Island fishers, the effect size for the municipality considered was estimated to be high. As opposed to this, when comparing distances travelled between Governor Generoso and Malita, the effect size for the municipality was estimated to be small.
These results were according to our original expectations and conform to results seen in images 7 and 8. Thus, we've obtained quantitative proof that fishers from Malita and Governor Generoso tend to fish farther from their port than fishers from Samal Island. Again, we believe this is due to them having a larger region of the gulf available to them, due to them possessing more advanced fishing vessels and due to them aiming for pelagic species via the use of hook and line fishing gears.
Distance ~ Time
Then, we tested the correlation between the distance vessels travelled and the month in which trips were recorded. Kruskal-Wallis analyses performed for this purpose never resulted in a p-value lower than 0,05.
As a result, we believe the fishers we've tracked tended to stick to the same general fishing areas across March, April and May. This was to be expected, as there were no management efforts or other events taking place during these months which we believed would influence the distribution of artisanal fishing efforts in the Davao Gulf.
We must note however, that our variances in this test were high, especially for our data from fishers of Malita (image 11). Presumably, this is due to an insufficient amount of data being available for each month. As a result, the test we've performed likely lack sufficient power for detecting a significant result even if there is one. Although power analyses for Kruskal-Wallis tests are possible, they were not performed in this Professional Practice due to the time and experience of the student being limited.
Speed Correlation Analyses
Along with a vessel's position and the direction it was headed, our GPS devices also logged the speeds at which vessels travelled. As a result, it was easy for us to analyze and compare the speeds at which vessels travelled for different municipalities, which informed us on the most frequently used types of fishing gears of a municipality and on the average engine power of their vessels.
In this section, we'll look at a "speed class frequency histogram", a figure often used for deducing a speed limit with which you can infer whether or not fishing vessels were actually fishing when a GPS point was recorded. Additionally, we'll discuss a correlation between recorded speeds and municipalities of origin.
Speed Class Frequency Histogram
An example of a speed class frequency histogram is shown in image 12. These histograms usually follow a general pattern, in which three peaks can be observed. Interesting is that each of these are associated with specific behaviors of fishers on sea.
A first peak is seen at 0 - 1 km/h. For the most part, this peak is contributed to by the immense amount of GPS recordings made when vessels were at 0 km/h. Scientists suggest that these recordings represent moments when fishers were either resting or performing fishing activities using static gears (such as a hook and line, image 9), but that it's okay to assume they only represent fishing activities during data analysis. (Lee et al., 2010)
A second peak is observed between 2 - 3 km/h. For the most part, this peak is usually contributed to by when vessels engage in fishing activities using low-speed, mobile gears (such as trawls). (Tingley et al., 2005)
GPS recordings at higher speeds are usually less frequent, as they're not associated with fishing activities. As a result, they're spread out over the tail-end of the histogram. In this latter portion however, a third peak is usually seen, which represents the average speed vessels tend to travel at when fishers move to- or from their fishing areas. (Forero et al. 2017)
Speed ~ Municipality
Our last result tested for a correlation between the speeds at which vessels travelled and the municipality they originated from.
This was done using the Kruskal-Wallis test, and the resulting p-value was again less than 2.2e-16. Thus, we proved there was a very significant correlation between the speed at which vessels travelled and their municipality of origin.
Additionally, we performed pairwise comparisons between each municipality. For this, p-values were corrected using the Bonferroni method. Levels of significance for these comparisons, along with estimations of their effect sizes, are shown in image 13.
Although we expected to see significant results between Governor Generoso and Samal Island, we didn't expect to see them in every comparison. Nonetheless, we believe these significant differences mainly arise from artisanal fishers of these municipalities possessing vessels capable of reaching and travelling at different speeds. This belief is also evidenced by the outliers seen in image 13, which correspond to the maximum speeds we recorded for vessels from these municipalities. Notable is that the maximum speed recorded for Samal Island exceeded that recorded from Malita, suggesting fishing vessels from both municipalities don't strongly differ in the maximum engine sizes seen there.
Also worth noting is the median value observed for fishers from Governor Generoso and Malita, which is at exactly 0 km/h. This is caused by the immense amount of recordings made at 0 km/h for these municipalities. As this isn't the case for Samal Island, we believe these fishers used static fishing gears less frequently.