Percidae Crystallaria asprella Etheostoma blennius E.
Sp. (Sunburst darter) E.
Variatum Percina antesella p.
Uranidea Cottidae Cottus aleuticus C.
Carolinae 252 Interval 3 Saddle 3 distance Saddled length Measurements were confined to the area of the body termed the saddled length (Fig. 2), defined as the length from the anterior border of the first saddle to the posterior border of the fourth saddle.
All species showed intraspecific variability in the width (distance from the anterior saddle margin to the posterior saddle margin) of the saddles; therefore, 60 Coitus carolinae Fig. 2.
Measurements used for the comparison of the saddled patterns of five species of North American fishes.
Boxed numbers are saddle numbers. 40 nation for its pattern.
It is unlikely that an organism utilizing saddles as disruptive coloration would be found normally against a uniform background because adding structure to an otherwise featureless background would make the organism more visible.
In fishes, the saddles may disrupt the body outline when the fish is viewed over a heterogeneous substrate such as gravel.
We examined several aspects of the saddle pattern and the ecological characteristics of North American fishes in order to determine what elements in the pattern are important and which form of cry psis the fishes are utilizing. — 20 40 60 80 Saddled length (mm) Fig. 3.
Linear regressions of saddle 2 distance and saddle 3 dis- tance on saddled length in five species of North American fishes. 253 Table 2.
Thkey tests for saddle 2 distance and saddle 3 distance.
Values greater than 0.05 (boldface) indicate that there is no significant difference between the two species in the placement of the saddle.
Saddle 2 distance C.
Uranidea N. [lavater b.
Saddle 3 distance C.
Uranidea N. [lavater C.
Uranidea N. [lavater 1.00000 0.00002 0.00002 0.07618 0.00002 1.00000 0.00002 0.08980 0.79586 0.00002 1.00000 0.12155 0.00002 0.00002 — 1.00000 0.00002 0.00002 0.00002 1.00000 0.01604 0.00002 1.00000 0.00002 1.00000 the exact positions of the saddles were defined as the midpoints of the saddles along the dorsal midline.
The midpoint of each saddle was determined by averaging the measurements taken at the anterior and posterior edges of the saddle.
Carolinae, the midpoint of the first saddle had to be estimated using the average width of the other three saddles because the posterior margin of the first saddle was usually indistinct on preserved specimens.
Patterns of the species were compared in two ways.
The first was a comparison between species of two saddle distances (Fig. 2): the distance between the anterior margin of the first saddle and the midpoint of saddle 2 (‘saddle 2 distance’), and the distance between the anterior margin of the first saddle and the midpoint of saddle 3 (‘saddle 3 dis- Table 3.
Tukey tests for comparison of intervals within species.
No comparisons were significant.
Degrees of freedom: N. [lavater =27, C.
Carolinae = 177, P.
Uranidea = 66, E.
Variatum = 177, E.
Blennius = 135.
Cottus carolinae Interval # 1 2 3 Percina uranidea Interval # 1 2 3 1 1.00000 2 0.00002 3 0.00002 Etheostoma blennius Interval # 1 1.00000 2 0.00002 3 0.00002 Etheostoma variatum Interval # 1 1.00000 0.00002 0.00002 1.00000 0.00002 1.00000 1.00000 2 0.00011 3 0.00011 Noturus [lavater Interval # 1.00000 0.04160 0.00076 1.00000 0.00011 — 0.0 C.
Uranidt a N.
FiavtJlef Species Fig. 4.
The relative spacing of all five species examined differed significantly from an evenly spaced pattern indicated by the line at 0.333.
The error bars indicate one standard deviation. tance’).
Saddle 4 was excluded from this comparison because the saddled length and the distance to the midpoint of saddle 4 were nearly equal, differing only by half of the width of saddle 4.
Ratios of the saddle 2 distance and saddle 3 distance to saddled length were determined for each individual and then arcsine transformed to normalize the values.
An ANCOVA was applied to the arcsine transformed measurements of each saddle separately, holding the log transformed saddled length as the covariate.
The ANCOVA was followed by Tukey tests in order to compare individual pairs of species.
The data were also plotted as a linear regression of saddle distance to saddled length (Fig. 3).
The second comparison determined whether the placement of the saddles on the body within a species deviated from an evenly spaced pattern.
An interval was defined as the distance between the midpoints of successive saddles (Fig. 2).
A ratio of each interval to the sum of the intervals was determined and arcsine transformed.
These intervals were chosen instead of measuring the interspaces directly because the sum of the ratios of the intervals equals 1, making it easier to compare the relative interval size both intraspecifically and interspecifically.
To test whether species deviated from an evenly spaced pattern, an ANOVA was performed for each species on the arcsine transformed ratios of interval lengths to saddled length.
The AN OVA’s were each followed by Tukey tests in order to determine if any of the following was true within a species: interval 1 = interval 2, interval 1 = interval 3, and interval 2 = interval 3. The significance of each of the linear regressions of saddle 2 distance and saddle 3 distance on saddled length (Fig. 3) was quite high for each of the five species (p < 0.00005).
All fishes, large or small, plotted closely to the regression line indicating that the effects of allometry are small.
The ANCOVA showed that there are differences between species in the exact position of the saddles, suggesting that exact placement of the saddles may not be very important in the evolution of the saddled pattern.
However, there are striking similarities among the species.
Tukey Tests (Table 3) indicate that the positions of saddle 2 and saddle 3 are not significantly different in C.
Carolinae and P uranidea (0.05 < P < 0.10 and 0.75 < P < 0.80), suggesting that the pattern is the same in the two species.
Saddle 2 in E.
Blennius is not significantly different from that of E.
Variatum (0.10 < p < 0.15), and the position of saddle 3 of E.
Variatum is not significantly different from that of C.
Carolinae (0.75 < P < 0.80).
In all species examined except N. [lavater, all four saddles are spaced unevenly (p < 0.001) (Fig. 4, Table 2) , and all intervals were significantly different from one another within a species (all p < 0.0005).
In N. [lavater, the pattern also was uneven (p < 0.001), but interval 2 was only slightly significantly larger than interval 1 (0.04 < P < 0.05).
With two intervals of nearly the same size, the pattern of N. [lavater tends more towards an evenly spaced pattern than that of the other four species in which interval 1 always was the largest and interval 3 was the smallest. Ecological characteristics of North American fishes with the saddle pattern To identify ecological characteristics of North American fishes with the saddle pattern, we used the information in Page (1983) and Page & Burr (1991).
We scored substrate and flow preferences as well as presence or absence of 3-5 dark dorsal saddles for all species of the families Catostomidae (61) and Ictaluridae (38), all darters (Percidae) (146), and all Cottus (Cottidae) (24).
Substrate was scored as either uniform (sand, mud , and bedrock) or un- 255 even (gravel, cobble, and boulders).
Flow was scored as no/lowflow (pools and lakes) or flow (riffles and runs).
The preference for flowing or standing water and the preference for uniform or uneven substrate were examined in 269 species, 49 of which have the saddle pattern.
All of the saddled species live on uneven, rocky substrates, and nearly all (N = 46) live in flowing water.
These results were analyzed using a Chi-square contingency table (Table 4) which showed that saddled species prefer uneven substrates and flowing water.
A Chi-square test was also applied to determine the relationship of flow to substrate.
A significant interaction between substrate type and flow (p < 0.0001) was found, as expected, because faster flowing water tends to wash out the smaller particles such as sand and mud while leaving the gravel, cobble, and boulders (Moyle & Cech 1988).
Because phylogenies are lacking for most groups of North American fishes, it is impossible to exclude the effects of phylogeny, and the Chi-square test probably suffers from pseudoreplication.
However, the data strongly suggest a positive relationship between living in flowing water over a rocky substrate and having the saddle pattern.
The saddle pattern is not primarily for obliterative countershading is the fact that the saddles of saddled species collected over sand are faded or lost.
Cottus carolinae, in particular, blanches to the color of the sand and the saddles are barely visible as slightly darker areas on the back.
If the saddle pattern served some function other than mimicking rocks, the saddles presumably would not fade when the fishes moved onto sand.
Also, pelagic fishes normally lacking saddles sometimes develop some form of saddle or bar pattern when they become benthic (Barlow 1963, Hailman 1982, Neil 1984).
This change often corresponds to a change in behavior from active movement to slow or no movement.
Finally, although juvenile hog suckers (Hypentelium spp.), which live in rocky riffles, are strongly saddled, hog suckers lose their saddle pattern as they grow and move into deeper, slower, less rocky habitats.
The five species examined, and many more species, have converged upon a pattern of four saddles.
In the five species examined, all except N [lavater possess a highly uneven pattern: interval 1 was the largest and interval 3 was the smallest.
This observation leads to three questions: 1.
What is the advantage of a pattern of unevenly spaced saddles? 2.
Why are the spaces between the saddles larger towards the head? 3.
Why doesn’t N [lavater conform to the pattern of the other four species? What is the advantage of a pattern of unevenly spaced saddles? When viewed from above (Fig. 5), each saddle delineates a light space that has the appearance of a rock, and a fish with four saddles is broken into a series of five ‘rocks’ (head, caudal fin, and three interspaces).
Animals maximize their crypsis by mimicking a random sample of the background (Endler 1978).
Because rocks in streams are a mixture of sizes and slopes, a fish that mimicked a series of rocks of similar proportions would be more conspicuous than one that mimicked a series of rocks of different sizes.
The five ‘rocks’ of different sizes on a foursaddled fish blend into the substrate.
Why are the spaces between the saddles larger towards the head? The body of a fish tapers from head to tail and, consequently, the widest space always will be the Discussion We propose that the selective advantage leading to the multiple evolution of the saddle pattern is through disruptive coloration.
The fishes with the saddle pattern are always found on gravel, and we propose that dark saddles mimic shadows or gaps between rocks while the light spaces between the saddles represent rocks.
Species with the saddle pattern also prefer flowing water.
However, three of the saddled species live in pools with little or no flow.
The pools in which these species live are rocky, and we assume that the main habitat element that has promoted the evolution of the saddle pattern is uneven, rocky substrates.
Flow may have aided in the evolution of the pattern because of additional shadows cast by surface ripples created by flowing water.
Additional evidence supporting the claim that 256 space nearest the head.
Because pieces of gravel in a stream are usually round or slightly elliptical, a fish can best mimic the substrate by having the longest spaces near the head.
Although the general pattern is for spaces to be longest near the head, C.
Carolinae has the ability to widen its first saddle such that the first space becomes smaller.
Carolinae can also go to the opposite extreme by eliminating its first saddle so that the first space is much larger and includes the head.
The ability to change the relative size of the saddles enhances the sculpin’s ability to camouflage itself.
A sculpin would be expected to eliminate the first saddle when it is among large gravel, but widen the first saddle when it is among smaller gravel.
Why doesn’t N.
Flavater conform to the pattern of the other four species? N.
Flavater deviates from the pattern exhibited by the other four species examined in that interval 1 and interval 2 are barely significantly different from one another.
It seems likely that, to a fish living on a rocky substrate, an uneven pattern is more cryptic than an even pattern; therefore, N.
Flavater must not rely on camouflage to avoid predation to the same extent as the other species examined.organisms under decreased predation pressure can afford to be less cryptic than organisms under high predation pressure (Endler 1978, 1986).
Madtoms reduce their predation pressure in two ways.
The first is that they Fig. 5.
Cottus carolinae in Big Creek Hardin Co., IL.
This individual shows the sculpin’s ability to widen its first saddle and, concomitantly, shrink the first space (photograph by Jonathan W.
Armbruster). Table 4.
Contingency table for the effects of flow (a) and of substrate (b) on the saddle pattern.
The Chi-square values indicate that saddled species are found more often in flowing water and on an uneven substrate than would be expected by chance.
Read more about Saddle 3 : carolinae 252 Interval 3 Saddle 3 distance Saddled length Measurements….: