A comparison of spatial learning and memory capabilities in intertidal gobies
Gemma E. White and Culum Brown
Department of Biological sciences, Macquarie University, North Ryde, NSW 2109, Australia.
Email: culum.brown@mq.edu.au
Supplementary Material: Tide Pool 4 analysis
Statistical analysis
A Kruskal-Wallis test was used to determine if there were differences in use of pool 4 during training
trials (mean of training trials 1-10) between the four goby species, followed by pairwise species contrasts using
Mann-Whitney U tests. To test if learning performance was affected by habitat preference, a Mann-Whitney U
test was used to compare which habitat (sand or rock pool) returned to the deepest pool (tide pool 1) most often.
Results
We found that sand species revisited tide pool 4 (shallow pool) to wait out low tide more often than
rock pool dwelling species during training trials (Mann-Whitney U: Z = -6.12, P <0.0001). This trend was also
observed after rock landmarks were moved for probe trials (Mann-Whitney U: Z = -6.201, P <0.0001)
(Supplementary Fig. 2).
All four goby species obtained significantly different use of tide pool 4 prior to probe trials (KruskalWallis: H = 38.95, df = 3, P <0.0001). Pairwise species contrasts using Mann-Whitney U tests revealed all
species were significantly different from one another in their learning performances (P <0.001 in all cases),
except for Cocos frillgoby and Krefft’s goby which did not differ from one another in learning performance (Z =
-0.897, P = 0.132) or eastern longfin goby and Hoese's sandgoby (Z = -2.17, P = 0.015) (Supplementary Table
1).
Highly significant differences were noted between rock pool and sand dwelling species contrasts, the
sand dwelling species (eastern longfin goby and Hoese's sandgoby) revisited tide pool 4 (shallow pool) ) to wait
out low tide more often than rock pool dwelling species (Cocos frillgoby and Krefft’s goby). Both sand dwelling
species were commonly left stranded in tide pool 4, while rock pool species made a few errors in the first few
trials the majority usually observed in tide pool 1 (deepest pool) especially by trial 8 (Supplementary Fig. 2).
Discussion
Tide pools 2-4 are very shallow by any fish’s standards. There is only just enough water at low tide in
pools 2-4 to keep the fish alive. This is not a comfortable spot for any fish. Thus there is no reason to suspect
that any fish would prefer any pool other than pool 1. Which is reflected somewhat in the results, no fish was
ever found in tide pool 2 or 3, however, they were found in tide pool 4, especially in the case of sand dwelling
species. Although sand dwelling species were observed in pool 3 as the tide was retreating (i.e. pool 1& 2 were
exposed while pool 3 & 4 are still under water) they quickly moved to pool 4 as tide pool 3 became
uncomfortably shallow, which at that stage was the deeper of the two pools available however it then continued
to drain leaving them stranded in tide pool 4. Which indicates that they were just remaining in the deepest part
of the tank (i.e. following the tides out) rather than actually learning the location of pool 4.
Supplementary Table 1 Mann-Whitney U pairwise species contrast results for prior rock movement (training
trial). Bonferroni corrected for multiple comparisons (i.e. to be significant P<0.008). Significant result is shown
in bold.
Species
No#
U-value
z-value
P-value
Cocos frillgoby (R) &
Krefft’s goby (R)
13
107.5
-1.18
0.132
Cocos frillgoby (R) &
Eastern longfin goby (S)
13
156
-4.24
<0.0001
Cocos frillgoby (R) &
13
182
-4.42
<0.0001
Hoese's sandgoby (S)
15
Krefft’s goby (R) &
13
156
-4.24
<0.0001
Eastern longfin goby (S)
12
Krefft’s goby (R) &
13
182
-4.42
<0.0001
Hoese's sandgoby (S)
15
Eastern longfin goby (S)
& Hoese's sandgoby (S)
12
182
-2.293
0.0079
13
12
15
Mean = mean percentage of trials fish waited out the low tide in pool 1 (deepest pool).
(R) = Rock pool dwelling species, (S) = Sand dwelling species.