Weijerman et al. Trends in biomass of coral reef fishes, derived from creel surveys in Guam 
241 
to extrapolate results of the interviews of a sample of 
fishermen from shore-based surveys to estimates of 
island-wide catch and effort (Bak 7 ). For comparisons of 
the fishery characteristics of the late 1980s with those 
of recent years, we used these WPacFIN estimates of 
aggregate effort and catch to examine possible shifts in 
fishing activities and catch composition. We somewhat 
arbitrarily chose to pool across 6-year periods to rep- 
resent the early and recent periods of fishery data. It 
was important to pool across multiple years to increase 
the amount of data available, especially for the recent 
period (2007-2012); data were pooled also because of 
large interannual variability. 
Fishery-independent surveys: calculation of biomass 
In 2011, scientists of the Coral Reef Ecosystem Program 
of the NOAA Pacific Islands Fisheries Science Center 
conducted an intensive, short-term stationary-point- 
count survey (133 sample sites) of shallow (depths <30 
m), hard-bottom coral reef areas around Guam. The 
methods used in this survey are described in detail in 
Williams et al. (2012) and briefly outlined here. Data 
of fish abundance and size distribution came from ran- 
dom visual surveys stratified into 3 depth ranges (<6 
m, 6-18 m, and 18-30 m). Because roving apex preda- 
tors, such as sharks (Carcharhinidae) and jacks (Ca- 
rangidae), are generally not well sampled by divers in 
small-area surveys, information on abundance and size 
distribution of species of roving apex predators came 
from broad-scale towed-diver surveys conducted around 
Guam in 2007, 2009, and 2011 by the Coral Reef Eco- 
system Program (Richards et al., 2011). 
Length estimates of fishes from visual censuses 
were converted to weight by using allometric length- 
weight conversion: 
W = a*TL b , (1) 
where a and b = constants; 
TL = total length in millimeters; and 
W = wet weight in grams. 
Length-weight parameters came from Taylor, 8 Taylor 
et al. (2012), Taylor and Choat (2014), FishBase (Froese 
and Pauly 9 ), and Nadon et al. (2015). In cases where 
length-weight information did not exist for a given 
species, parameters from congeners were used. 
For each taxon, trophic classification was based on 
diet information, largely from FishBase. Using biomass 
density from the diver surveys and known areas of 
habitat from GIS maps, which were adapted within the 
Coral Reef Ecosystem Program from other GIS prod- 
ucts (NCCOS 10 ), we estimated biomass per functional 
8 Taylor, B. 2012. Personal commun. James Cook Univer- 
sity, Townsville, Queensland 4811, Australia. 
9 Froese, R., and D. Pauly (eds.). 2015. FishBase, vers. 
10/2015. [World Wide Web electronic publication; available 
at website, accessed March 2015.] 
10 NCCOS (National Centers for Coastal Ocean Science). 
2005. National Centers for Coastal Ocean Science, Shallow- 
group and for all fishes combined; these estimates were 
minimum values because cryptic and nocturnal fishes 
generally are undercounted by daytime visual surveys. 
Estimation of catchability and reconstruction of historical 
fish biomass 
A reconstructed time series of reef-fish biomass can be 
estimated from a CPUE time series and gear-specific 
catchability coefficients (Haddon, 2001). Generally, it 
is assumed that catch rates are linearly related to 
stock biomass and that the catchability coefficient is 
constant (Haddon, 2001). For the reconstruction in our 
study, we estimated the relationship between CPUE (as 
a proxy for relative biomass) and fish population bio- 
mass by incorporating a fishery-independent estimate 
of biomass in 2011, as described in the previous sec- 
tion. This approach requires that the CPUE data come 
from a representative sample, where the catch was 
taken in a consistent way by one or more fishing meth- 
ods. The CPUE data used in this study had limitations, 
namely that the CPUE data for taxa infrequently en- 
countered by any gear type in the DAWR creel surveys 
were not reliable; infrequently encountered taxa in- 
cluded many reef fishes, although the CPUE data have 
been shown to be reliable for some of the jacks (Bak 
Hospital 11 ). Because of these limitations, in addressing 
federal management by means of annual catch limits, 
the CPUE time series available from the creel surveys 
were rejected for reef-fish stock assessments (Sabater 
and Kleiber, 2014). However, in this study, we were in- 
terested not in absolute abundance or stock size but 
in temporal trends of fish populations and relative dif- 
ferences between the late 1980s and recent years, and 
these data are suitable for that purpose. 
In the shore-based surveys, 9 gear types were dif- 
ferentiated (Table 2). The gear type hooks-and-gaffs 
was used almost entirely to catch octopus (98% of 
hook-and-gaff landings for the period 1985-2012), and 
the type other methods included gears used in glean- 
ing for invertebrates and algae (60% and 16% of total 
landings per respective gear type; WPacFIN 5 ) (Hens- 
ley and Sherwood, 1993). Therefore, for the historical 
reconstruction of reef-fish biomass, we excluded catch 
and effort data for the gear types hooks-and-gaffs and 
other methods from analyses. We also excluded catch 
and effort data for the gear type cast nets from esti- 
mation of reconstructed reef-fish biomass because cast 
nets were used primarily to catch juvenile fishes (i.e., 
rabbitfishes [Siganus spp.], goatfishes [Mullidae], and 
Water Benthic Habitats of American Samoa, Guam, and the 
Commonwealth of the Northern Mariana Islands (CD-ROM). 
Silver Spring, MD. [Metadata available at website] 
n Bak Hospital, S. 2015. Western Pacific creel survey pro- 
gram data summary and analysis: Guam, the Common- 
wealth of the Northern Mariana Islands, and American Sa- 
moa. NOAA Pac. Islands Fish. Sci. Cent. Admin. Rep. H- 
15-06C, 194 p. [Available from Pac. Islands Fish. Sci. Cent., 
Natl. Mar. Fish. Serv., NOAA, 1845 Wasp Blvd., Bldg. 176, 
Honolulu, HI 96818.] 
