Kim Hanson, Ryan Ray, and Austin Reed
Cornell College
Dept. of Biology
Mt. Vernon, Iowa
January 2002

Introduction

Two species of bromeliads, Tillandsia utriculata and T. fascicuata, have recently been named to Florida’s endangered species list due to predation by Metamasius callizona, the Mexican weevil (Frank, 2002). Patterns of predation are currently being monitored by Teresa Cooper, who has been monitoring the condition of individual T. utriculata and T. fasciculata host plants in Myakka River State Park (MRSP) since August, 2001. In collaboration with Cooper, we mapped additional populations of T. utriculata and T. fasciculata and tested hypotheses about their distribution patterns with MRSP.

Tillandsia utriculata and T. fasciculata are wind dispersed. Their seeds germinate only on the bark of trees (Bennett, 1989; Benzing, 2000). Those two factors, wind dispersal and bark dependence, lead us to believe that T. utriculata and T. fasciculata will be found most commonly on whichever tree type has the highest density in the habitat, as these trees provide the most potential germination sites.

Microenvironmental factors affect the suitability of the germination site for T. utriculata and T. fasciculate seedlings. Bark stability, texture, porosity, and chemical composition affect the ability of bromeliad seedlings to take hold of a tree (Bennett, 1987). Roots may take up to an entire month to form (Benzing, 1980). Once formed, the roots enter the bark of the tree but never the vascular tissue (Benzing, 2000). It is easier for epiphytes to take root on rough surfaces as opposed to on smooth, loose, and exfoliating bark surfaces, which provide a poor home for a T. utriculata or T. fasciculata seed in even the most perfect environmental conditions for growth (Benzing, 1980; Bennett, 1987; Bennett, 1992). Based on these factors, we predicted we would find T. utriculata and T. fasciculata most commonly on trees, with multiple branches and rough bark structure and less commonly on smooth non-branching palm.

Tillandsia, with over 550 identified members, is the largest genus in Bromeliaceae (Benzing, 2000; Dimmitt, 1992). Tillandsia members can tolerate a wide range of temperatures from 32-100º, but have a low tolerance for heavy shade (Benzing, 2000; Dimmitt, 1992). Members of the genus can be found living in mangroves, cypress swamps, and pine flat wood forests, but are most commonly found living in live oak hammocks, which are typically nutrient poor (Bennett, 1989). Tillandsias get a majority of their nutrients from water absorbed through the canopy (Benzing, 2000; Frank 2001). Water is absorbed through a leaf surface densely covered by peltate trichomes (Sugden, 1981). Members of the Tillandsia genus take several years to mature and even more time to become robust seeding adults (Benzing, 2000).

The seed structure of both T. utriculata and T. fasciculata is very similar. Their seeds have coma that are "sticky" to the bark of certain types of trees (Benzing, 2000). The coma exhibit bifurcate cross walls that favor adhesion (Benzing, 2000). Tillandsia utriculata and T. fasciculata rely on the coma of their seedlings for attachment to a germination site, as well as for increased distance of seed dispersal (Benzing 2000; Bennett, 1989; Bennett, 1992).

Because T. utriculata and T. fasciculata are two of largest species of Tillandsia in southern Florida, they have become the prime target for the Mexican weevil, which cannot complete its larval development in smaller species (Benzing, 1981). The weevil lays eggs down inside the bromeliad. The larval form of the weevil then mines down into and eats the meristematic tissue, killing the host plant (Frank, 2002). Tillandsia utriculata is the larger of the two species and tends to be more rounded at the base. Tillandsia. utriculata can take up to ten years to seed and will only seed once before dying (Bennett, 1989). The white inflorescence of T. utriculata is usually less branched than that of T. fasciculata and has a mean seed production of 13,272 (Bennett, 1989). Tillandsia fasciculata has a bright red inflorescence that is more highly branched than its close relative T. utriculata. T. fasciculata will seed multiple times in its life time and can flower all year around, although most commonly in the summer and spring (Frank, 2001).

Myakka River State Park (MRSP) is located in Sarasota & Manatee Counties in Florida. MRSP has a variety of habitats including: wetlands, sloughs, marsh, forests, and prairies. We chose three study sites that extended Cooper’s research to areas of the park previously unmapped and monitored. Deer Prairie Slough (DPS) is an area that is saturated for most of the year and has a dense mixture of oak and palm trees. Ranch House Road (RHR) is a flood plain surrounded by prairie and is moderately dense with large oak and palm. Horse Corral (HC) is a palmetto prairie containing a sparse mixture of oak, palm and pine trees (Figure 1).

We used Cooper’s (2002) method of tiered structure for sampling T. utriculata and T. fasciculata. Tier I areas are very large sections of land (e.g. southern Florida) which contain smaller units called Tier II areas (e.g. drainage basins in south Florida). Within tier II areas are tier III areas (e.g. state parks or preserves), which may or may not cross political borders. Tier III areas are then broken down into tier IV areas, which include between ten and one hundred host plant areas. A host plant area (HPA) is defined as a tree or a patch on the ground that contains host plants. Host plant areas can contain multiple host plants. Tier IV areas are then broken down into at least three equally divided sections, each containing between three to ten host plants areas. Half of the sections within the tier IV are then randomly chosen for mapping and monitoring. A tier V area would refer to a specific host plant area, and a tier VI area would refer to a specific host plant. Cooper (2002) maps and sketches both tier V and tier VI areas so they can be found and monitored in the future.

To test our hypothesis that T. utriculata and T. fasciculata are less likely to be found on palm trees than on other species of trees, we surveyed host plant areas (tier V) within randomly sampled sections of five tier IV areas. The host plant areas were surveyed to determine the total number and location of T. utriculata and T. fasciculata. Host plant areas, as well as landmark trees were mapped using a base line measurement and triangulation methods with a compass. At all sites, we noted the location of T. utriculata and T. fasciculata in host plant areas. If the bromeliads were on trees, we noted whether they were attached to the trunk or on a branch. If the bromeliads were on the ground, we noted whether they were still attached to twigs or not. Also noted were the types of trees the host plants were found on, either palm, deciduous, or pine. To determine the area of land we sampled trees in, we took a base line measurement of the path and then counted all host plants fully visible from either side of the path. The area of our samples sampled varied slightly due to differing tree density, which made visibility different in each habitat.

Among the census sites in MRSP: oak, palm, and pine all supported T. fasciculata and T. utriculata. Within the three tier IV’s found at Deer Prairie Slough (DPS) the density of oak was .562 trees per foot (t/ft). The density of palm found in this was .177 t/ft. Bromeliad density was .427 t/ft throughout the habitat. Of the 388 bromeliads found in DPS only 2 were found on palms (Table 1).

Horse Corral (HC) tier IV had an oak density of .205 d/ft, a palm density of .125 d/ft, and a pine density of .089 d/ft. The bromeliad density was .026 d/ft. Of the 13 bromeliads found in this tier IV, 12 were located on oaks and 1 was found on a pine (Table 2).

Ranch House Road (RHR) tier IV had an oak density of .133 d/ft and a palm density of .401 d/ft. The bromeliad density was .066 d/ft. Of the 63 bromeliads found in RHR 43 were found on oaks and 10 were on palm (Table 3).

The total T. fasciculata and T. utriculata count for the three habitats censused in MRSP yielded 454 bromeliads. Of those, 97% were found on, in, or around oak trees and 2.6% were found on palms. Less than 1% of T. fasciculata and T. utriculata were found on pines, which represent 3% of trees (Figure 2).

We hypothesized that the trees with the highest density in a given sample area, would act as hosts to T. fasciculata and T. utriculata. We found that Deer Prairie Slough had the highest overall density of trees as well as the highest overall density of bromeliads. The oak to palm ratio was .562:.177 but, over 99% of the bromeliads were found on oaks. If bromeliad seeds were dispersed randomly without preference we would have expected more bromeliads to be found on palms. We used a chi square test to determine if distribution of bromeliads were random with respect to tree type. It was expected that 79% of the trees with bromeliads would be oak, and 21% palm. However, it was observed that 95% of the trees containing bromeliads were oak, and 5% palm. This difference is statistically significant (C ²=6.1; p<0.01{Table 4}).

Horse Corral had the lowest overall density of trees as well as the lowest overall density of bromeliads. This is consistent with the hypothesis that the fewer the trees that occupy an area the fewer bromeliads will be found. This habitat was sparsely made up of oaks, palms, and pines, with a ratio of .205:.125:.089 respectively. Out of the 13 bromeliads found in this habitat, 12 were located on oak, 1 was located on a pine and none were found on palms. This is consistent with the hypothesis, and is due to the fact that oaks have the highest density in this area. Although the pines had the lowest density in this sample, the fact that one bromeliad was found in a pine can be explained due to the spatial area occupied by the pines.

For HC we expected that 51% of the trees with T. fasciculata and T. utriculata would be oak, 13% would be palm, and 36% would be pine. Instead it was observed that 92% of the trees containing bromeliads were oak, 8% were pine, and palms had zero. These differences are shown to be statistically significant (C ²=8.78; p<0.05 {Table 5}). If the numbers of bromeliads had followed the distribution of trees there would have been bromeliads on all three types of trees. However, there were no bromeliads located on palms, and only one on a pine tree. These numbers are not simply due to chance alone.

Ranch House Road was the only site sampled where the density of palms was greater than oaks. The ratio of oaks/palms/pines was .133/.401/.004 respectively. Out of the 53 bromeliads found in this area, 43 were located on oak trees, 10 on palms, and were absent from pine. Although the density of oaks in this sample was much lower then that of palms, the oaks were still the primary hosts to the bromeliads. This is most likely due to the large surface area available for germination on oaks. The number of bromeliads found on palms in this area was much higher than other areas due to the high density of palms. Because pines were rare in this habitat compared to those of oak and palm, it is not surprising that bromeliads were not found on pines.

For RHR we expected that 30% of the trees with T. fasciculata and T. utriculata would be oak, 69% palm and 1% pine if dispersion were random with respect to tree type. However we observed that 75% of the trees containing bromeliads were oak and 25% palm, with an absence from pines. This was statistically significant (C ²=11.67; p<0.01{Table 6}). Although there were very few pines on RHR, it was still numerically predicted that bromeliads would have been present. It was also expected that since palm trees made up most of the tree density, there should have been a large number of bromeliads on the palms. Instead, the largest numbers were on oaks.

A cumulative expected value for trees MRSP containing T. fasciculata and T. utriculata was determined to be 65% oak, 32% palm, and 3% pine. It was observed that 90% of trees containing T. fasciculata and T. utriculata were oak, 8% palm, and 2% were pine. These numbers are also statistically significant (C ²=20.1; p<0.001{Table 7}). The percentage of oak observed was much greater than that expected, whereas that of palm was smaller. This shows that values throughout MRSP are not simply due to chance alone.

With the exception of RHR, the tree type with the highest density in a given habitat acted as the primary host for bromeliads. At the RHR site we found palms to have the highest density, but did not act as the primary host. However, the proportion of bromeliads found on palms in this area was much greater when compared to the other sample sites. This may be explained due to the fact that palm trees can only offer a single trunk as a germination site for the seedlings. Oak trees, on the other hand, have multiple branches per trunk which offers much more surface area for the seedlings to attach to.

Bennett, Bradley. 1989. "The Florida Bromeliads: Tillandsia utriculata". Journal of the Bromeliad Society 39:265-271.

Bennett, Bradley. 1992. "The Florida Bromeliads: Guzmania monostachia". Journal of the Bromeliad Society 42:266-270

Bennett, Bradley. 1987. "Spatial distribution of Catopsis and Guzmania ‘Bromeliaceae’ in southern Florida". Bulletin of the Torrey Botanical Club 114:265-271.

Benzing, D. 2000. Bromeliaceae: Profile of an Adaptive Radiation. Cambridge University Press.

Benzing, D. 1980. The Biology of the Bromeliads. Mad River Press Inc.

Benzing, D. 1981. "Bark Surfaces and the Origin and Maintenance of Diversity Among Angiosperm Epiphytes: A Hypothesis". Selbyana. 5:248-255

Cooper, Teresa. 2001.

Dimmitt, Mark. 1992. Bromeliads: A Cultural Manual. The Bromeliad Society, Inc.

Frank, H. (2001). Save Florida’s Native Bromeliads . Retrieved January 7, 2002, from University of Florida website: http://entomology.ifas.ufl.edu/frank/savebromeliads/index.htm

Sugden, Andrew. 1981. "Aspects of the Ecology of Vascular Epiphytes in two Colombian Cloud Forests. II. Habitat Preferences of Bromeliaceae in the Serrania De Macuria". Selbyana 5:264-273.