44 



Fishery Bulletin 100(1) 



Figure 2 



Transverse section of a six-year-old Mayan cichlid tCichlasuma urophthalmiis) otolith 

 showing the outer region (ORl, inner region ( IRi, and five visible annuli ( 1-51. Note that 

 the first annulus (1) corresponds to the fish's second year of growth. A ring correspond- 

 ing to the first year of growth was not consistently visible and was therefore not counted. 

 Measurements for marginal-increment analysis were made on an axis adjacent to the 

 sulcal ridge from the core (C) to the dorsolateral margin (DLM). Scale bar=500 /im. 



on each otolith section was counted independently by two 

 readers using compound microscopes, and the results were 

 compared. If there was a discrepancy in the counts be- 

 tween readers, the section was re-examined. If a consensus 

 could not be reached between the readers after the third 

 reading, the otolith was excluded from the study. 



Linear regi-ession was used to determine the relation- 

 ship of otolith radius to standard length and marginal- 

 increment analysis was used to determine the periodicity 

 of ring formation. Distance from the core to the proximal 

 edge of each ring and to the dorsolateral margin of the oto- 

 lith (otolith radius) was measured (Fig. 2). Measurements 

 were made with a digital-image processing system along 

 an axis adjacent to the sulcal ridge. The distance from 

 the outermost ring to the dorso-lateral margin (i.e. mar- 

 ginal increment=MI) was plotted by month (marginal in- 

 crement analysis). Because the majority of Mayan cichlids 

 in Taylor River spawn during May and June (Faunce and 

 Lorenz, 2000), and ring formation occurred during Janu- 

 ary-May, we assigned each fish a biologically realistic me- 

 dian hatching date of 1 June. Fish collected prior to 1 June 

 that had not yet formed a new opaque ring (=high MI), 

 and all fish collected after 1 June, were assigned a yearly 

 age equal to their ring count. Fish collected before 1 June 

 that had already formed a new opaque ring (i.e. an "early" 

 ring) were assigned a yearly age of one less than their ring 



count. To compare the timing of ring formation between 

 age groups, marginal-increment analysis was performed 

 on pooled ages 0-3 and 4-7 because our monthly sample 

 sizes for individual age classes were insufficient for this 

 analysis. 



We used linear regression to determine the relationship 

 between standard length and total length for all hook- 

 and-line caught fish. The relationship between standard 

 length and total weight was calculated separately for each 

 sex with logjy-transformed data. Analysis of covariance 

 (ANCOVA) was used to test for significant differences be- 

 tween the slopes and intercepts of male and female length- 

 weight relationships. Length-frequency distributions for 

 males and females caught with hook-and-line were com- 

 pared by using the Mann-Whitney rank sum t-test. Non- 

 linear least squares procedures (SAS, 1989) were per- 

 formed on final obsen'ed age-at-length data to estimate 

 parameters for the von Bertalanffy gi'owth equation 



L, =L..(l-exp\-Kit-t„)]), 



where L, = the standard length (mm); 

 L = the asymptotic length; 

 K = the Brody growth coefficient; 

 t = the age (years); and 

 tfy = the age at zero length (von Bertalanffy, 19.57). 



