PMC:4264897 / 31767-40780 JSONTXT

Annnotations TAB JSON ListView MergeView

{"target":"https://pubannotation.org/docs/sourcedb/PMC/sourceid/4264897","sourcedb":"PMC","sourceid":"4264897","source_url":"https://www.ncbi.nlm.nih.gov/pmc/4264897","text":"Comparison of condition index methods\nWe used three approaches (, R i, and ANCOVA) to analyze the effect of two factors (i.e., sex and diet) on body condition in a previously published data set (Kelly and Tawes 2013). These methods produced results that would lead to very different biological interpretations of how sex and diet affect body condition in G. texensis crickets at eclosion (Table3; Fig.2). suggests that neither sex nor diet affect body condition, whereas R i, suggests that males are in significantly better condition than females with diet having a marginally non-significant effect on condition. In contrast, the use of ANCOVA to statistically control for differences in body size among crickets suggests that individuals on a good diet are in significantly better condition than those reared on a poor diet with sex having little effect on body condition.\nTable 3 Analysis of effect of sex and diet quality on body condition in Gryllus texensis using three different condition index methods. Data are from Kelly and Tawes (2013). Bold type indicates a significant factor effect at α  = 0.05\nFactor Scaled mass index () Ordinary least squares residual index (R i) ANCOVA\nF (df) P -value F (df) P -value F (df) P -value\nSex 2.008 (1,171) 0.158 5.364 (1,171)* 0.022 1.29 (1,170)† 0.258\nDiet 1.276 (1,171) 0.260 2.959 (1,171) 0.087 132.4 (1,170)† \u003c0.001\n* P  \u003c 0.05 for the Levene's homoscedasticity test.\n† No heterogeneity of slopes in the ANCOVA method at α  = 0.05.\nFigure 2 Effect of sex and diet quality (good quality = grey boxes; poor quality = white boxes) on body condition index in Gryllus texensis as estimated by three different methods for the Kelly and Tawes (2013) data. Boxplots are shown for the (A) scaled mass index () and (B) residual index (R i). Boxes represent the lower (25%) and upper (75%) quartiles, the solid dark horizontal line represents the median, and the whiskers indicate 1.5 times the interquartile range. Data beyond the end of the whiskers are outliers and plotted as black dots. (C) Mean (±1 SE) body mass (log g) from ANCOVA after adjusting for body size (log mm; circles = good diet; triangles = poor diet). Samples sizes are: low-quality females, n  = 35; low-quality males, n  = 40; high-quality females, n  = 47; high-quality males, n  = 52. See Table3 for details of statistical tests. We found that SMA better describes the scaling relationship between body mass and body size than OLS regression whether we use the SMA slope from the reference population (2.642 ± 0.11 vs. 2.011 ± 0.088; z  = 4.54, P  \u003c   0.001) or from the Kelly and Tawes (2013) data set (2.32 ± 0.088 vs. 2.011 ± 0.088; z  = 2.46, P  =   0.014). Using the SMA slope derived from the Kelly and Tawes (2013) data set (i.e., b SMA = 2.32) produced results that were very similar (sex: F  =   2.57, df = 1,171, P  =   0.111; diet: F  =   0.001, df = 1,171, P  =   0.980) to those using the slope from the reference population (i.e., b SMA = 2.642). This might not always be the case, however, and so we recommend that biologists use b SMA from our reference populations in their calculations of in G. texensis or A. domesticus.\nPrior to calculating R i using a common slope (b OLS) with the Kelly and Tawes (2013) data set, we tested whether the slopes of the groups within each treatment (i.e., males vs. females within “sex” and poor vs. good diets within “diet”) differed significantly (see Garcia-Berthou 2001). The OLS slopes did not differ between the sexes (males vs. females: b OLS = 2.16 ± 0.14 vs. b OLS = 1.92 ± 0.11, t  =   1.33, df = 170, P  =   0.185) or diets (poor vs. good: b OLS = 2.05 ± 0.14 vs. b OLS = 182 ± 0.14, t  =   1.16, df = 170, P  =   0.25). We thus pooled the data and calculated R i for each individual using a common slope (b OLS = 2.01 ± 0.088). Residual body mass was then used to examine the effects of sex and diet on body condition.\nTesting the homogeneity of slopes assumption is not the only assumption that requires examination, however, because the elevation of the regression lines also plays an important role in potentially biasing residual calculations and scaling coefficients (Garcia-Berthou 2001). For example, despite there not being an interaction between the sex-specific OLS slopes in the Kelly and Tawes (2013) data set, we found that the male slope had a significantly greater intercept (i.e., elevation) than the female slope (sex: t  =   2.30, df = 171, P  =   0.023). That is, after removing the non-significant interaction and then testing for differences between the sexes in adjusted body mass while statistically controlling for body size (i.e., performing an ANCOVA), we found that the adjusted mean body mass of males was significantly greater than that for females (Fig.3). Thus, our common slope had a lower elevation than the male-specific slope and a higher elevation than the female-specific slope. Ignoring this fact has serious implications for the calculation of residual body mass because the common slope produced residuals that were spuriously large for males and small for females. In contrast, the intercepts (i.e., adjusted means) did not significantly differ between poor and good diets (diet: t  =   1.852, df = 171, P  =   0.066). Consequently, our statistical analysis using R i as the condition index suggested that there is a significant difference between the sexes, with diet quality having little effect on condition.\nFigure 3 Relationship between body mass (ln g) and body size (ln mm) as described by ordinary least squares (OLS) regression in Gryllus texensis for (A) each sex separately (males: triangles and dashed line; females: circles and dotted line) and (B) for each level of diet quality (poor diet: triangles and dashed line; good diet: circles and dotted line) for the Kelly and Tawes (2013) data. The common OLS slope is represented by the solid line in both plots. Using R i is a well-accepted and established approach in the field of evolutionary ecology despite having several drawbacks (Kotiaho 1999; Garcia-Berthou 2001; Green 2001; Peig and Green 2009, 2010). Our re-analysis of Kelly and Tawes’ (2013) data highlights that OLS regression does not accurately describe the scaling relationship between body mass and body size and that ignoring key statistical assumptions leads to spurious differences between treatment factors.\nFinally, the significant effect of diet using ANCOVA appears to be a result of this method simply tracking the significant differences in body mass and size between the diet treatments (Fig.4). Crickets that were reared on a good quality diet were larger in body mass (F  =   37.07, df = 1,172, P  \u003c   0.0001) and size (F  =   38.28, df = 1,172, P  \u003c   0.0001) than those on a low-quality diet. The sexes did not differ significantly in the size of either trait (body mass: F  =   0.036, df = 1,172, P  =   0.85; body size: F  =   0.80, df = 1,172, P  =   0.37), and consequently, ANCOVA showed no effect of sex on body condition. These findings are similar to those of Peig and Green (2010) who also found that ANCOVA produced significant differences in condition according to differences in body size in several vertebrate taxa.\nFigure 4 Boxplots showing differences in body mass (g) and body size (pronotum length, mm) between the sexes and levels of diet quality in Gryllus texensis for the Kelly and Tawes (2013) data. The box represents the lower (25%) and upper (75%) quartiles, the solid dark horizontal line is the median, and the whiskers indicate 1.5 times the interquartile range. Data beyond the end of the whiskers are outliers and plotted as black dots (females, n  = 82; males, n  = 92; high-quality diet, n  = 99; low-quality diet, n  = 75). P -values are from one-way ANOVAs testing for differences in mass or size between each treatment level (see text). In conclusion, we showed that the scaling relationship between different body mass components and different measurement of body size varies between species and between the sexes with a species. We strongly recommend that biologists not assume that a particular index is a reliable indicator of body condition but rather they empirically verify the reliability of the condition index. We also show that the best description of the scaling relationship between body mass and body size in both cricket species was produced by SMA, rather than OLS, regression and our values for b SMA from our reference populations should be used by biologists when calculating in G. texensis and A. domesticus. Finally, our re-analysis of the data presented in Kelly and Tawes (2013) illustrates well the dangers of analyzing body condition using methods that are based on least squares in the y -plane. Both R i and ANCOVA suggested that body condition was significantly affected by a different treatment factor while suggested no treatment effects. Thus, depending on the method employed, very different biological conclusions would be drawn from the same data.","divisions":[{"label":"title","span":{"begin":0,"end":37}},{"label":"p","span":{"begin":38,"end":874}},{"label":"table-wrap","span":{"begin":875,"end":1509}},{"label":"label","span":{"begin":875,"end":882}},{"label":"caption","span":{"begin":884,"end":1110}},{"label":"p","span":{"begin":884,"end":1110}},{"label":"table","span":{"begin":1111,"end":1391}},{"label":"tr","span":{"begin":1111,"end":1192}},{"label":"th","span":{"begin":1111,"end":1117}},{"label":"th","span":{"begin":1119,"end":1139}},{"label":"th","span":{"begin":1141,"end":1184}},{"label":"th","span":{"begin":1186,"end":1192}},{"label":"tr","span":{"begin":1193,"end":1245}},{"label":"th","span":{"begin":1193,"end":1199}},{"label":"th","span":{"begin":1201,"end":1209}},{"label":"th","span":{"begin":1211,"end":1217}},{"label":"th","span":{"begin":1219,"end":1227}},{"label":"th","span":{"begin":1229,"end":1235}},{"label":"th","span":{"begin":1237,"end":1245}},{"label":"tr","span":{"begin":1246,"end":1317}},{"label":"td","span":{"begin":1246,"end":1249}},{"label":"td","span":{"begin":1251,"end":1264}},{"label":"td","span":{"begin":1266,"end":1271}},{"label":"td","span":{"begin":1273,"end":1287}},{"label":"td","span":{"begin":1289,"end":1295}},{"label":"td","span":{"begin":1297,"end":1310}},{"label":"td","span":{"begin":1312,"end":1317}},{"label":"tr","span":{"begin":1318,"end":1391}},{"label":"td","span":{"begin":1318,"end":1322}},{"label":"td","span":{"begin":1324,"end":1337}},{"label":"td","span":{"begin":1339,"end":1344}},{"label":"td","span":{"begin":1346,"end":1359}},{"label":"td","span":{"begin":1361,"end":1366}},{"label":"td","span":{"begin":1368,"end":1382}},{"label":"td","span":{"begin":1384,"end":1391}},{"label":"table-wrap-foot","span":{"begin":1392,"end":1509}},{"label":"footnote","span":{"begin":1392,"end":1444}},{"label":"label","span":{"begin":1392,"end":1393}},{"label":"p","span":{"begin":1395,"end":1444}},{"label":"footnote","span":{"begin":1445,"end":1509}},{"label":"label","span":{"begin":1445,"end":1446}},{"label":"p","span":{"begin":1448,"end":1509}},{"label":"figure","span":{"begin":1510,"end":2372}},{"label":"label","span":{"begin":1510,"end":1518}},{"label":"caption","span":{"begin":1520,"end":2372}},{"label":"p","span":{"begin":1520,"end":2372}},{"label":"p","span":{"begin":2373,"end":3183}},{"label":"p","span":{"begin":3184,"end":3926}},{"label":"p","span":{"begin":3927,"end":5460}},{"label":"figure","span":{"begin":5461,"end":5923}},{"label":"label","span":{"begin":5461,"end":5469}},{"label":"caption","span":{"begin":5471,"end":5923}},{"label":"p","span":{"begin":5471,"end":5923}},{"label":"p","span":{"begin":5924,"end":6391}},{"label":"p","span":{"begin":6392,"end":7222}},{"label":"figure","span":{"begin":7223,"end":7866}},{"label":"label","span":{"begin":7223,"end":7231}},{"label":"caption","span":{"begin":7233,"end":7866}},{"label":"p","span":{"begin":7233,"end":7866}}],"tracks":[{"project":"MyTest","denotations":[{"id":"25512844-23637808-26483245","span":{"begin":211,"end":215},"obj":"23637808"},{"id":"25512844-23637808-26483246","span":{"begin":1043,"end":1047},"obj":"23637808"},{"id":"25512844-23637808-26483247","span":{"begin":1716,"end":1720},"obj":"23637808"},{"id":"25512844-23637808-26483248","span":{"begin":2631,"end":2635},"obj":"23637808"},{"id":"25512844-23637808-26483249","span":{"begin":2759,"end":2763},"obj":"23637808"},{"id":"25512844-23637808-26483250","span":{"begin":3264,"end":3268},"obj":"23637808"},{"id":"25512844-23637808-26483251","span":{"begin":4315,"end":4319},"obj":"23637808"},{"id":"25512844-23637808-26483252","span":{"begin":5843,"end":5847},"obj":"23637808"},{"id":"25512844-23637808-26483253","span":{"begin":6161,"end":6165},"obj":"23637808"},{"id":"25512844-23637808-26483254","span":{"begin":7405,"end":7409},"obj":"23637808"},{"id":"25512844-23637808-26483255","span":{"begin":8627,"end":8631},"obj":"23637808"}],"namespaces":[{"prefix":"_base","uri":"https://www.uniprot.org/uniprot/testbase"},{"prefix":"UniProtKB","uri":"https://www.uniprot.org/uniprot/"},{"prefix":"uniprot","uri":"https://www.uniprot.org/uniprotkb/"}],"attributes":[{"subj":"25512844-23637808-26483245","pred":"source","obj":"MyTest"},{"subj":"25512844-23637808-26483246","pred":"source","obj":"MyTest"},{"subj":"25512844-23637808-26483247","pred":"source","obj":"MyTest"},{"subj":"25512844-23637808-26483248","pred":"source","obj":"MyTest"},{"subj":"25512844-23637808-26483249","pred":"source","obj":"MyTest"},{"subj":"25512844-23637808-26483250","pred":"source","obj":"MyTest"},{"subj":"25512844-23637808-26483251","pred":"source","obj":"MyTest"},{"subj":"25512844-23637808-26483252","pred":"source","obj":"MyTest"},{"subj":"25512844-23637808-26483253","pred":"source","obj":"MyTest"},{"subj":"25512844-23637808-26483254","pred":"source","obj":"MyTest"},{"subj":"25512844-23637808-26483255","pred":"source","obj":"MyTest"}]},{"project":"2_test","denotations":[{"id":"25512844-23637808-26483245","span":{"begin":211,"end":215},"obj":"23637808"},{"id":"25512844-23637808-26483246","span":{"begin":1043,"end":1047},"obj":"23637808"},{"id":"25512844-23637808-26483247","span":{"begin":1716,"end":1720},"obj":"23637808"},{"id":"25512844-23637808-26483248","span":{"begin":2631,"end":2635},"obj":"23637808"},{"id":"25512844-23637808-26483249","span":{"begin":2759,"end":2763},"obj":"23637808"},{"id":"25512844-23637808-26483250","span":{"begin":3264,"end":3268},"obj":"23637808"},{"id":"25512844-23637808-26483251","span":{"begin":4315,"end":4319},"obj":"23637808"},{"id":"25512844-23637808-26483252","span":{"begin":5843,"end":5847},"obj":"23637808"},{"id":"25512844-23637808-26483253","span":{"begin":6161,"end":6165},"obj":"23637808"},{"id":"25512844-23637808-26483254","span":{"begin":7405,"end":7409},"obj":"23637808"},{"id":"25512844-23637808-26483255","span":{"begin":8627,"end":8631},"obj":"23637808"}],"attributes":[{"subj":"25512844-23637808-26483245","pred":"source","obj":"2_test"},{"subj":"25512844-23637808-26483246","pred":"source","obj":"2_test"},{"subj":"25512844-23637808-26483247","pred":"source","obj":"2_test"},{"subj":"25512844-23637808-26483248","pred":"source","obj":"2_test"},{"subj":"25512844-23637808-26483249","pred":"source","obj":"2_test"},{"subj":"25512844-23637808-26483250","pred":"source","obj":"2_test"},{"subj":"25512844-23637808-26483251","pred":"source","obj":"2_test"},{"subj":"25512844-23637808-26483252","pred":"source","obj":"2_test"},{"subj":"25512844-23637808-26483253","pred":"source","obj":"2_test"},{"subj":"25512844-23637808-26483254","pred":"source","obj":"2_test"},{"subj":"25512844-23637808-26483255","pred":"source","obj":"2_test"}]}],"config":{"attribute types":[{"pred":"source","value type":"selection","values":[{"id":"MyTest","color":"#ec93cc","default":true},{"id":"2_test","color":"#93e6ec"}]}]}}