Viz Lab Summer Grant 2007

2008 Reports 2007 Reports 2006 Reports 2005 Reports 2004 Reports

Avian Foraging Mechanisms in Relation to Preferences in Goldenrod Gall Predation

Terry Hams Graduate Student, IBS

Background and methods

The foraging of resident birds in Minnesota is subject to the extremes of the seasons. The birds must therefore find sources of food to cope with these stresses. One method that these birds have discovered is to prey upon insect galls. Downy woodpeckers and Black capped chickadees are resident birds that must seek out sources of protein in the winter to supplement their diets. The chickadees are mostly insectivorous during the summer months and shift their diet to half seeds and half protein during the winter. The woodpeckers are primarily insectivores in the summer and remain that way in the winter as well. These diets require that the birds seek out insect larva and one such source is that of galls.

The goldenrod gallfly is a gall forming insect that is widespread throughout North America. It oviposits eggs into the stems of goldenrod plants in the early summer where they form a spherical gall over the next few months. During the fall the insect will create an exit tunnel to the edge of the gall and then will go into a suspended animation to survive the winter, and will emerge in the late spring as an adult fly. The gallfly must contend with two insect predators in order to survive until the spring. The Eurytoma wasp is a parasitic wasp that attacks goldenrod galls in the summer months and oviposits eggs into the gallfly larva. The second predator is a mordellid beetle that lays it eggs on the surface of a gall where the larvae emerge and bore into the gall. These larvae then tunnel throughout the gall digesting the plant tissue and will consume any other larva that they encounter, such as the gallfly or wasp.

To better understand how these birds choose the galls that they attack during the winter I formulated four hypotheses. The first hypothesis is that birds will prefer to attack the largest galls there by maximizing there reward. The second hypothesis is birds will prefer prairie galls over forest galls. Hypothesis three is the birds will prefer galls that contain both mordellid beetles and gallflies to that of gallflies alone. That last hypothesis is birds will avoid attacking galls that have no exit tunnel present.

To test these hypotheses I created gall stands that were a 2x4 base stand with 100 three foot bamboo stakes placed vertically along it. On these stakes were randomly placed 50 prairie and 50 forest galls all with a diameter between 20 and 25mm. The concentration of mordellid beetle has been shown to be much higher in the prairie galls then the forest galls which is the rational for using these two types. The size range would allow for a small chance of the parasitic wasp to be present as they are size limited in attacking gall by the length of their ovipositor.

Gall stands were placed in natural patches of Goldenrod that had already been predated to assure that the birds were near the area. The sites used were; Hawk Ridge Bird Observatory, Hartley Nature Area, Jay Cooke state park, Lester golf course, UMD research farm, and Enger golf course. Each site had two replicates placed at least 250meters apart. The stands were placed in the field on the second week of January 2007, within three meters of the edge of trees or bushes that provide cover for birds in their foraging. The stands were then surveyed once a week to record any predation by birds or damage from other sources. Once 25% of the galls in a stand were attacked the entire stand was removed and the galls dissected to determine contents. Stands that were not attacked were removed in the middle of April before they warmed up and allow the insects to emerge. The dissections would allow me to determine the factors that the birds were selecting for in choosing to attack these galls such as insect inhabitant, gall diameter, and presence of an exit tunnel. Additionally, I collected galls from natural patches of Goldenrod in some of these areas to observe if the same selection factors were being used by the birds.

Results

The first hypothesis that birds will prefer to attack larger galls was accepted. Analysis of the data allowed us to observe that in the natural patches of goldenrod, the predated galls were significantly larger then mean gall size. When this was organized by the specific avian predator it was observed that the Black capped chickadees attacked the largest galls available while the Downy woodpeckers did not select for large galls (Figure 1). The second hypothesis that birds prefer prairie galls over forest galls was rejected. The birds showed significant selection for forest galls overall. The woodpeckers had a much stronger selection for the forest galls then the chickadees (Figure 2).

Hypothesis three that birds will prefer galls containing both mordellid beetles and gallflies was rejected. The birds showed selection against galls containing mordellid beetles (Figure 3), which may explain the reason that prairie galls were avoided as well. The woodpeckers were more selective then the chickadees in avoiding mordellid beetles. The final hypothesis that birds will prefer galls with exit tunnels was accepted. The woodpeckers strongly selected for this factor while the chickadee selection was weaker yet still significant (Figure 4).

In conclusion the avian predators showed significant selection for many of the tested factors. The woodpeckers and chickadees were selecting for similar factors yet in most cases the woodpecker’s selection was stronger then the chickadee’s. This may be because the woodpecker has developed senses that allow it to better determine the contents of a goldenrod gall before it expends the energy to excavate it. Chickadees on the other hand may be just selecting the largest galls available in order to increase their chance of finding a gallfly larva to consume. Although excavating larger galls is more costly on the bird it is still likely more profitable then excavating many small galls for the same biomass reward.

Figure 1 Comparison of size of mean gall size of galls attacked by Black capped chickadees and Downy woodpeckers. The mean gall size is represented by the dashed line. Chickadees showed significant (P<0.05) selection for larger galls.

Figure 2 The selection for biome by the avian predators. Both predators showed significant selection for forest gall (P<0.05) but the woodpecker selection was much stronger.

Figure 3 Preference of avian predators for the presence of Mordellid beetles in galls. The woodpecker showed strong selection against the mordellids (P<0.01). The dashed line represents the mean abundance of mordellids in all galls.

Figure 4 Preference of avian predators for the presence of an exit tunnel. Both predators showed significance for exit tunnels (P<0.05). The dashed line represents the mean abundance of exit tunnels present in the galls.

Modeling

The Viz Lab (Visualization and Digital Imaging Lab) issued me a grant for the summer of 2007 in order to conduct modeling of the avian predations I observed in my experiment. The modeling will be done with the use of the Forager 1.1 program by Amber Waves Software. This program has five foraging models incorporated within that the data from these experiments can be inserted into. The five foraging models are; dynamic programming, hierarchical matching model, marginal value theorem, marginal value theorem with learning, and the predator-prey model.

Of these models I will be using the dynamic programming and hierarchical matching models. The dynamic programming model allows the user to enter data on prey abundance in a patch as well as time spent foraging in a patch. Using this data I can then estimate a fitness response of the predators from the observed results of my experiment and adjust the model to see how the fitness changes. The other model that will be used is the hierarchical matching model. In this model I can enter data on the abundance and palatability of the three different prey items in each patch to observe the habitat use and time spent foraging in the patches. This will allow me to model both the avian predators and prey items and determine the foraging differences.

There are two additions experiments that Viz Lab is providing support for. The first is to quantify the avian predation of galls by verifying avian attack through digital recordings. These recordings will help me to determine exactly which gall was attacked by which avian predator in order to give me more accuracy in the foraging models. At this point determining which bird attacked a gall is subjective in that the chickadee is believed to have a larger attack area then the woodpecker. This has not been tested to assure accuracy and by recording which bird attacked which gall I will be able to observe patterns that each predator shows in its attack.

Figure 5 The traditional method for determining attacked by avian predator. The size of the attack as thus far been used to assign an either a chickadee or woodpecker as the attacker. http://www.facstaff.bucknell.edu/abrahmsn/solidago/gallkey8.html

The second experiment that is to be conducted is to determine a method for locating exit tunnels in galls without having to dissect them. The woodpecker and possibly the chickadee are able to do this through an undetermined sensory mechanism. I plan on using ultraviolet and infrared light photography to see if the exit tunnels can be observed in this manner. Another possible method may be using small microphones and computer programs to observe the vibration signature of galls to find a pattern for contents.