Category: Group Project
Participants: Rebecca Marks, Jherimechee Womack, Millie Chamblee, April McCrae
Site: Palo Verde

Keywords: Erythroxylon havanense, herbivory, leaf miner, altitudinal gradient

Summary:

Herbivoric activity upon Erythroxylon havanense along the Cerros Calizos was compared to changes in altitude.  While numerically this was illustrated, divergent observations may show other factors to be established at a later time.
 

Introduction:

Plant predation rates can be studied and monitored in an attempt to discover plant/predator relationships and to recognize notable adaptations in native environments.  Primary producers can experience diverse levels of herbivory within a species.  These varying levels can be because of differences in defense chemical production while individual differences are often attributed to leaf age or changes caused by herbivoric activity (Lindquist 1997).  Abiotic factors that can induce these variations could include such things as temperature, humidity, levels of sun penetration, and altitude.

In studying Erythroxylon havanense predation, knowledge of tree growth, occurrence and availability is essential.  Erythroxylon havanense is a shrub tree that grows in the dry forest understory.  The leaves appear as small elliptical shapes with a rounded apex (Gentry 1993).  Within the Palo Verde reserve, mature Erythroxylon appears as a relatively common plant, occurring at a population of greater than 10 individuals per hectare.  The small treelets, ranging in size from <10 cm up to 5 meters, can be found in abundance on slopes with at least a 5% gradient and experiencing good drainage (Janzen 1983).
 
Upon first examination of the species, it was noted that individuals growing at the base of the Mirador appeared to experience little predation by indeterminate leaf miners.  Conversely, plants growing in mid-elevation (150-300 ft from base) appeared to suffer greater predation.  While plant specimens at the peak (400 ft from base) appeared to suffer lesser predation.  It was this group’s hypothesis that leaf miner predation of Erythroxylon havanense was directly related to the altitudinal occurrence of the plant.  It was also held that, in opposition to the null hypothesis offered in Laura Lundquist’s paper, that the distribution of attack on any given leaf was not random.
 

Methods:
 
The group studied Erythroxylon specimens found along the Cerros Callizos path leading to the limestone cliff above the OTS station.  The first three plots were marked at 150 foot intervals beginning at the base of the trail.  The fourth plot was stationed at the Mirador Guayacan overlook measuring 400 feet of elevation.  At each site three trees were selected within a 15 foot radius and were marked for specimen collection.  Fifty leaves were randomly taken from each tree.  Air temperature, ground temperature and relative humidity were recorded in close proximity of each tree.  These readings were recorded at noon, and early evening of July 19, 1999 and early morning of July 20, 1999.  Percent leaf cover approximations were also recorded at each site.
 
Leaf examinations were conducted by measuring leaf length, damage to the leaf, number of predation sites per leaf, and size of each predation site.  When quantified using spreadsheet software, the following data was revealed: average number of leaves affected by miner predation, percentage of damage to random leaves from each tree, average number of predation sites on each affected leaf, average size of predation sites at each plot level, average relative humidity at each plot, average ground and air temperature at each plot.
 
To determine random dispersal of larvae within each leaf 25 leaves with greater than or equal to 3 predation sites were collected.  The leaves were tested using a variance of the Poisson Distribution Test.  A graph was created (known as the Womack Quadrant Graph) on a transparency in order to measure the location of the miner site in a quadrant fashion.  Each leaf was graphed using the center of the leaf as the zero point and its petiole aligned with the X axis (see fig.1).  Once all data was collected the numbers were compared to a random numbers chart based on a four-quadrant distribution chart.  Once these numbers were compared a Chi square test was done to determine the probability of randomness.
 

Results:
 
Our findings revealed that when plotting the average occurrence of predators, the original observations were valid.  The leaves that were prominent with miner damage were located in plots two and three at 150 foot and 300 foot levels than in plots one and four located at 100 and 400 foot levels.  It was noted that sun penetration appeared to be greater for plot four.  In regards to temperature, plot one showed an average of 81 degrees, plot two showed an average of 81 degrees, plot three showed an average of 83 degrees and plot four showed an average of 83 degrees.  Average humidity was 81, 83, 75, and 74 respectively for plot one, two, three, and four.  The average miner damage of leaves at each plot was 30% (plot 1), 58% (plot 2), 57% (plot 3), 13% (plot 4).
 

Discussion:
 
Numerical representation of our original observations appear to show that indeed, Erythroxylon havanense predation occurs more at mid-altitude on the Mirador trail then it does at low and high altitudes.  However, it can also be noted that low altitude specimens on the trail is near forest edge as is peak altitude.  Taking altitude and forest growth patterns into account, the general statement that middle region trees experience greater predation seems valid.  Based on numerical evidence, the proposed null hypothesis that there is no difference in the average number of leaf miner sites per leaf between individual [trees] (Lundquist 1997) is not so.  At mid-altitude/mid-forest levels, it was observed that, among damaged leaves, the average number of spots was greater then it was at base and peak sites.  It was also noted that smaller specimens seemed to experience more predation then the larger specimens.
 
After observing ground temperature, air temperature, and relative humidity there appeared to be no correlation.  This led to the belief that sun availability or soil composition may have an effect on individual plants at each site.  While soil composition could not be studied due to time constraints, sun availability estimations were performed.  The percent canopy cover was estimated at each site.  It was observed that at site one there was ~75% cover, site two showed ~60% cover, site three had ~ 40% cover and site four showed the least with ~5% coverage.  The original thought that increased sun availability led to decreased predation was proven to be incorrect as site one had little sunlight and little predation.  However, L. Lundquist's summary of the carbon/nitrogen balance hypothesis stated that increased light availability should lead to increased photosynthesis.  This should increase the production of carbon based defensive chemicals, while an area with significantly less sunlight should produce greater amounts of nitrogen based chemical defenses (Lundquist 1997).  This concept leads to an interesting observation.  It has been modeled that in site one, the plants are growing in an area of ~75% shade.  If Lundquist's summary is correct then one would expect a greater production of nitrogen based defenses as opposed to carbon based defenses.  At site two, with an approximate 60% shade, and site three with an approximate 40% shade, would cause a close to equal level of nitrogen and carbon based defenses.  Site four, having ~95% sunlight availability would have an increased carbon based defense.  It appears that plants with equal production of carbon and nitrogen based defenses are more prone to attack than plants with an increased level of one or the other.  While this has not been proven, it does pose an interesting question.  Do chemical defense production levels impact predation levels, and does light availability affect production enough to produce this phenomenon?
 
Lundquist’s previous observation that on certain individuals the occurrence of leaf miners was not distributed randomly appears to be correct.  It was observed that the vast majority of mining sites were not found near the edges of leaves.  In addition, those leaves having more than one area of predation rarely demonstrated overlapping mine sites.  Upon investigation of randomly selected leaves, the distribution of miners appeared to be intentional.  Using a modified distribution graph to compare larval mine site dispersal to a random numbers table, it became apparent that the arrangement of sites was not random.  It was also found that, when graphed, leaves with three or more mine sites had an average of 10 mm between each mine.  This regularly occurring pattern also indicates that the mine sites are not randomly placed on the leaf.  Further research will lead to a better understanding of the placement of mine dispersal on the leaves of Erythroxylon havanense.
 

Acknowledgments:
 
Dr. Ted Stiles and Dr. Phillipe Hensel were instrumental in providing extensive knowledge and support for our research.  Without their expertise, our research would not have been as successful or as much fun.
 

Literature Cited:

Lundquist, L.  1997. “Occurrence of leaf miners on leaves of Erythroxylon havanense”.  OTS Tropical Biology, 97-1:1- 3.

Gentry, A. H.  1993.  “A Field Guide to the Families and Genera of Woody Plants of Northwest South America”.  The University of Chicago/London.

Janzen, D. H.  1983.  “Costa Rican Natural History”.  University of Chicago Press, Chicago/London.