NOTE Nolan and Danilowicz: Use of crest nets for sampling presettlement larvae of reef fishes in the Caribbean Sea 
217 
Table 1 (continued) 
Summer 
Spring 
Summer 
2005 
2006 
2006 
Family 
Genus 
Crest Channel 
Species net net 
Crest Channel 
net net 
Crest Channel 
net net Total 
Pomacentridae 
Stegastes 
adustus 
3 
0 
0 
0 
1 
0 
4 
(continued) 
Stegastes 
part it us 
2 
1 
0 
0 
23 
1 
27 
Striped Stegastes 
All 
171 
2 
0 
1 
258 
12 
439 
Scaridae 
Sparisoma 
spp. 
329 
3 
359 
3 
838 
0 
1532 
Scorpaenidae 
Scorpaena 
spp. 
44 
2 
202 
1 
64 
2 
315 
Serranidae 
Diplectrum 
spp. 
45 
5 
0 
0 
5 
0 
55 
Pseudogramma 
gregoryi 
95 
0 
38 
0 
117 
0 
250 
Rypticus 
sp. 
2 
0 
0 
0 
5 
0 
7 
Hypoplectrus 
spp. 
0 
0 
0 
0 
7 
0 
7 
Sphyraenidae 
Sphyraena 
barracuda 
125 
13 
6 
0 
39 
6 
189 
Syngnathidae 
Cosmocampus 
spp. 
341 
1 
111 
0 
95 
0 
548 
Tetraodontidae 
Sphoeroides 
spp. 
0 
0 
7 
3 
16 
0 
26 
Canthigaster 
spp. 
435 
0 
24 
7 
26 
0 
492 
(when 2005 data were analyzed separately, a very simi- 
lar plot was obtained). Most species were captured in 
greater abundance with crest nets and rarely, if ever, 
caught in the channel nets. For example, the families 
Acanthuridae, Ogcocephalidae, and Pomacanthidae were 
only caught in crest nets and there were no species 
or families that were exclusively caught in channel 
nets. The summer and spring sampling periods were 
extremely different (Fig. 2). However, when the three 
sampling periods were plotted separately, very similar 
ordinations with respect to environmental factors were 
obtained. The difference between summer and spring 
in the combined ordination of Figure 2 could be due to 
the lower numbers of larvae captured in spring 2006; 
however, there were notable absences of families in that 
sampling period, e.g., no Chaetodontidae or Ogcocephali- 
dae and only a single representative of Pomacentridae. 
Of the environmental variables (Fig. 2), the onshore 
wind was positively correlated with abundance and spe- 
cies richness of larval reef fishes sampled in crest nets. 
The combined factor (nocturnal illumination and tidal 
periodicity) was important but did not align strongly 
with the other explanatory or species variables. Higher 
water temperatures at the net sites corresponded with 
fewer larvae caught because water temperature was 
negatively correlated with the presence of the vast ma- 
jority of species. 
Time series analyses 
Peaks and lows in the supply of fish larvae appeared 
on the same nights in reef crest nets and channel nets 
in the mangroves (Fig. 3). The cross-correlation plots 
between net types revealed that catches (both in terms 
of abundance and species richness) were significantly 
correlated at a lag of zero (data sets were aligned for 
correlation on the same day at a lag of zero, one data 
set leads the other by one day for correlation at a lag 
of +1, etc.). For abundance, the greatest correlation 
between net types was at a lag of zero days (Fig. 3A). A 
lesser correlation at a lag of plus three days indicates 
that some groups of larvae took three days to pass from 
the reef crest to the mangrove channels. The other sig- 
nificant correlations at lags of -4, -3, and -1 days are 
more difficult to explain. There seems to be no biologi- 
cal reason that cohorts of reef fish larvae should arrive 
in the mangrove channels up to four days before they 
arrive at the reef crest. This finding may be a result of 
pooling abundances of all species and could possibly be 
resolved with further analysis by splitting abundances 
into families or species (where possible). Species rich- 
ness was also correlated at a lag of zero days; however, 
the other significant correlation, at a lag of -4 days, 
was greater than that at day zero (Fig. 3B). As with 
abundance, there seems to be no biological explanation 
for this correlation and more detailed analysis may prove 
advantageous. 
The auto-correlation plot for abundance (Fig. 4A) il- 
lustrates that there was no periodicity in the flow-cor- 
rected data and that the catch on any one night was 
not correlated with that on the preceding or following 
nights. However, the plot for species richness (Fig. 4B) 
shows a lunar periodicity in the numbers of species 
caught. The significant negative correlation at a lag of 
16 days (at just over half the lunar cycle) shows that 
greater numbers of species caught in new-moon periods 
