Crabtree et al.: Feeding habits o f Albula vulpes 
755 
have resulted from the seagrass die-off, these changes 
could potentially affect both feeding and occurrence 
of bonefish in Florida Bay. In this article, we describe 
the feeding habits of bonefish from waters off the 
Florida Keys. We consider both length-related and sea- 
sonal changes in bonefish diet. In addition, we com- 
pare the diets of bonefish collected from two important 
Keys areas: Florida Bay (including parts of Everglades 
National Park and adjacent waters) and ocean-side 
(Florida Straits) fishing areas off the Florida Keys. 
Methods 
Collections 
We examined stomach contents of 385 bonefish col- 
lected from South Florida waters from December 
1991 to April 1995. Most of these bonefish were 
caught with hook-and-line gear in waters off the 
Florida Keys, Florida Bay, and Biscayne Bay during 
daylight hours either by biologists or by a single pro- 
fessional bonefish guide and his anglers. Supplemen- 
tal collections of small bonefish (<425 mm FL, n- 22) 
were made with seines and gill nets of various sizes 
in waters off the Florida Keys. Bonefish were placed 
on ice immediately after capture. Fork length (FL) 
was later measured to the nearest millimeter (mm), 
stomachs were removed, and the contents preserved 
in 10% buffered formalin. Contents of individual 
stomachs were sorted and identified to the lowest 
possible taxon. Fragments of prey organisms were 
counted as one, unless countable parts such as eye 
lenses were found. Weights of prey organisms were 
measured by blotting prey items on filter paper and 
weighing them on an analytical balance. The num- 
ber of individuals of each food type as a percentage 
of the total number of identifiable prey items (per- 
cent numerical abundance, N), the percentage of 
stomachs containing prey in which a particular prey 
taxa occurred (frequency of occurrence, F), and wet 
weight as a percentage of the total weight of all prey 
items (percent weight, W) were determined. For the 
larger and more abundant prey taxa (alpheid shrimp, 
penaeid shrimp, portunid crabs, xanthid crabs, and 
Opsanus beta), we measured prey size to examine 
the relation between predator and prey size. We 
measured total length (TL, tip of the rostrum to tip 
of the uropod) of shrimp, carapace width of crabs, 
and standard length (SL) of O. beta. 
1 Matheson, R. E., D. A. Camp, S. M. Sogard, and K. A. Bjorgo. 
1998. Changes in seagrass-associated fish and crustacean com- 
munities on Florida Bay mud banks: the effects of recent eco- 
system changes? Manuscript in review. 
We compared the abundance of prey found in bone- 
fish stomachs with the abundance of benthic and 
epibenthic crustaceans and fishes from typical 
Florida Keys bonefish habitat. Information on the 
abundance of potential prey in Florida Bay is based 
on meter-square throw-trap collections by Matheson 
et al. 1 during 1994-96. We used data for Buchanan 
Bank (n=30 collections) in the Atlantic suben- 
vironment as described by Zieman et al. (1989) and 
followed by Matheson et al. 1 The Atlantic suben- 
vironment, and specifically the Buchanan Bank area 
sampled by Matheson et al., 1 is an area where many 
of our Florida Bay bonefish were caught. Data on 
prey abundance from ocean-side (Florida Straits) 
areas of the Florida Keys are from 54 samples that 
we collected following the methods of Sogard et al. 
(1987) and Matheson et al. 1 with meter-square throw 
traps. Ocean-side samples were taken during Sep- 
tember 1996 (n= 14) and January 1997 (n=40) at vari- 
ous locations from the middle Keys north to Elliot 
Key. We sampled areas where we had previously 
caught bonefish and that appeared to be representa- 
tive of typical ocean-side bonefish flats. Throw-trap 
samples were collected over a different time period 
(1994-97) than that for our bonefish specimens 
( 1991-95), and we assumed for our comparisons that 
prey availability did not change over this time. If prey 
abundance changed during 1991-97, this change 
could have biased our comparisons. 
Data analysis 
Nonparametric multivariate techniques were used 
to analyze feeding data. Similarity matrices were 
constructed with pairwise Bray-Curtis similarity 
coefficients (Bray and Curtis, 1957). Square-root- 
transformed, percent-standardized prey-weight data 
were used to generate similarities. Prey weight was 
used for all comparisons except feeding selectivity 
comparisons, because this measure more closely re- 
flects the energetic importance of a prey species in 
the diet than does either frequency of occurrence or 
percent numerical abundance. Percent numerical 
abundance was used in feeding selectivity compari- 
sons because we were interested in the relative abun- 
dance of prey in stomachs and in the environment. 
Hierarchical agglomerative cluster analysis that in- 
corporated a group-average linking method was used 
to search for groups among bonefish stomach con- 
tents. A nonparametric ordination technique, 
nonmetric multidimensional scaling (MDS), was used 
to ordinate sites on the basis of the similarity ma- 
trix. The contribution of the various prey categories 
to the percentage similarity within groups and the 
differences among groups were estimated with a simi- 
