Limits...
When field experiments yield unexpected results: lessons learned from measuring selection in White Sands lizards.

Hardwick KM, Harmon LJ, Hardwick SD, Rosenblum EB - PLoS ONE (2015)

Bottom Line: Color did not have a significant effect on survival, but we found several unexpected relationships including variation in predation over small spatial and temporal scales.In addition, we detected a marginally significant interaction between sex and color, suggesting selection for substrate matching may be stronger for males than females.We use our results as a case study to examine six major challenges frequently encountered in field-based studies of natural selection, and suggest that insight into the complexities of selection often results when experiments turn out differently than expected.

View Article: PubMed Central - PubMed

Affiliation: Department of Biological Sciences, University of Idaho, Moscow, Idaho, United States of America.

ABSTRACT
Determining the adaptive significance of phenotypic traits is key for understanding evolution and diversification in natural populations. However, evolutionary biologists have an incomplete understanding of how specific traits affect fitness in most populations. The White Sands system provides an opportunity to study the adaptive significance of traits in an experimental context. Blanched color evolved recently in three species of lizards inhabiting the gypsum dunes of White Sands and is likely an adaptation to avoid predation. To determine whether there is a relationship between color and susceptibility to predation in White Sands lizards, we conducted enclosure experiments, quantifying survivorship of Holbrookia maculate exhibiting substrate-matched and substrate-mismatched phenotypes. Lizards in our study experienced strong predation. Color did not have a significant effect on survival, but we found several unexpected relationships including variation in predation over small spatial and temporal scales. In addition, we detected a marginally significant interaction between sex and color, suggesting selection for substrate matching may be stronger for males than females. We use our results as a case study to examine six major challenges frequently encountered in field-based studies of natural selection, and suggest that insight into the complexities of selection often results when experiments turn out differently than expected.

Show MeSH
Proportion of lizards that survived trials with respect to year and enclosure location.Survivorship differed significantly between years (P = 0.03) and among enclosures (P < 0.01) in our general linear model. In addition, we detected an interaction between year and enclosure (P = 0.03), where survivorship in some enclosures differed significantly between years and survivorship in other enclosures did not. Dashed lines represent the interaction effect between year and enclosure on survivorship that we detected in our linear model, with endpoints representing mean proportion of survivors in replicates within different enclosures in different years. Solid, vertical lines indicate the standard error of the mean for enclosures in years with multiple replicates, where survivorship varied among replicates. We used open replicate data from each year to calculate survivorship for each enclosure (one replicate in 2011 and four in 2012 for enclosure A [n = 70 lizards]; one in 2011 and two in 2012 for enclosures B and C [n = 42 lizards each]; zero in 2011 and four in 2012 for enclosure D [n = 56 lizards]).
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4340912&req=5

pone.0118560.g003: Proportion of lizards that survived trials with respect to year and enclosure location.Survivorship differed significantly between years (P = 0.03) and among enclosures (P < 0.01) in our general linear model. In addition, we detected an interaction between year and enclosure (P = 0.03), where survivorship in some enclosures differed significantly between years and survivorship in other enclosures did not. Dashed lines represent the interaction effect between year and enclosure on survivorship that we detected in our linear model, with endpoints representing mean proportion of survivors in replicates within different enclosures in different years. Solid, vertical lines indicate the standard error of the mean for enclosures in years with multiple replicates, where survivorship varied among replicates. We used open replicate data from each year to calculate survivorship for each enclosure (one replicate in 2011 and four in 2012 for enclosure A [n = 70 lizards]; one in 2011 and two in 2012 for enclosures B and C [n = 42 lizards each]; zero in 2011 and four in 2012 for enclosure D [n = 56 lizards]).

Mentions: The results of our general linear model indicated that lizard survivorship did not vary with respect to paint treatment (χ21 = 0.037, P = 0.85), but did vary with respect to enclosure location and year. Survivorship was significantly higher in trials in 2012 compared with trials in 2011 (χ21 = 4.46, P = 0.03) (Fig. 3). In addition, survivorship for trials in 2011 and 2012 differed dramatically among enclosures (χ23 = 37.58, P < 0.01); two enclosures exhibited higher survivorship with 67% (95% CI [5%, 100%]) and 86% (95% CI [62%, 100%]) of lizards recaptured, and two enclosures exhibited lower survivorship with 41% (95% CI [11%, 71%]) and 31% (95% CI [11%, 51%]) of lizards recaptured (Fig. 3). Our model detected a significant interaction between enclosure and year (χ22 = 7.11, P = 0.03), where survivorship in some enclosures differed significantly between years and survivorship in other enclosures did not (Fig. 3). We also observed a marginally significant interaction between paint treatment and sex (χ21 = 3.64, P = 0.05). Mean survivorship of mismatched males was 48% (95% CI [28%, 68%]), and survivorship of matched males was 58% (95% CI [38%, 78%]). The opposite pattern occurred in females, with 67% (95% CI [53%, 81%]) of mismatched individuals surviving, compared with 50% (95% CI [27%, 73%]) of matched individuals (Fig. 4). However, posthoc tests comparing survivorship between treatments for each sex individually were not statistically significant (Fisher’s exact tests, P = 0.23 for males and P = 0.12 for females). Finally, lizard size (i.e., snout-vent length) did not have a significant effect on survivorship (χ21 = 2.94, P = 0.09).


When field experiments yield unexpected results: lessons learned from measuring selection in White Sands lizards.

Hardwick KM, Harmon LJ, Hardwick SD, Rosenblum EB - PLoS ONE (2015)

Proportion of lizards that survived trials with respect to year and enclosure location.Survivorship differed significantly between years (P = 0.03) and among enclosures (P < 0.01) in our general linear model. In addition, we detected an interaction between year and enclosure (P = 0.03), where survivorship in some enclosures differed significantly between years and survivorship in other enclosures did not. Dashed lines represent the interaction effect between year and enclosure on survivorship that we detected in our linear model, with endpoints representing mean proportion of survivors in replicates within different enclosures in different years. Solid, vertical lines indicate the standard error of the mean for enclosures in years with multiple replicates, where survivorship varied among replicates. We used open replicate data from each year to calculate survivorship for each enclosure (one replicate in 2011 and four in 2012 for enclosure A [n = 70 lizards]; one in 2011 and two in 2012 for enclosures B and C [n = 42 lizards each]; zero in 2011 and four in 2012 for enclosure D [n = 56 lizards]).
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4340912&req=5

pone.0118560.g003: Proportion of lizards that survived trials with respect to year and enclosure location.Survivorship differed significantly between years (P = 0.03) and among enclosures (P < 0.01) in our general linear model. In addition, we detected an interaction between year and enclosure (P = 0.03), where survivorship in some enclosures differed significantly between years and survivorship in other enclosures did not. Dashed lines represent the interaction effect between year and enclosure on survivorship that we detected in our linear model, with endpoints representing mean proportion of survivors in replicates within different enclosures in different years. Solid, vertical lines indicate the standard error of the mean for enclosures in years with multiple replicates, where survivorship varied among replicates. We used open replicate data from each year to calculate survivorship for each enclosure (one replicate in 2011 and four in 2012 for enclosure A [n = 70 lizards]; one in 2011 and two in 2012 for enclosures B and C [n = 42 lizards each]; zero in 2011 and four in 2012 for enclosure D [n = 56 lizards]).
Mentions: The results of our general linear model indicated that lizard survivorship did not vary with respect to paint treatment (χ21 = 0.037, P = 0.85), but did vary with respect to enclosure location and year. Survivorship was significantly higher in trials in 2012 compared with trials in 2011 (χ21 = 4.46, P = 0.03) (Fig. 3). In addition, survivorship for trials in 2011 and 2012 differed dramatically among enclosures (χ23 = 37.58, P < 0.01); two enclosures exhibited higher survivorship with 67% (95% CI [5%, 100%]) and 86% (95% CI [62%, 100%]) of lizards recaptured, and two enclosures exhibited lower survivorship with 41% (95% CI [11%, 71%]) and 31% (95% CI [11%, 51%]) of lizards recaptured (Fig. 3). Our model detected a significant interaction between enclosure and year (χ22 = 7.11, P = 0.03), where survivorship in some enclosures differed significantly between years and survivorship in other enclosures did not (Fig. 3). We also observed a marginally significant interaction between paint treatment and sex (χ21 = 3.64, P = 0.05). Mean survivorship of mismatched males was 48% (95% CI [28%, 68%]), and survivorship of matched males was 58% (95% CI [38%, 78%]). The opposite pattern occurred in females, with 67% (95% CI [53%, 81%]) of mismatched individuals surviving, compared with 50% (95% CI [27%, 73%]) of matched individuals (Fig. 4). However, posthoc tests comparing survivorship between treatments for each sex individually were not statistically significant (Fisher’s exact tests, P = 0.23 for males and P = 0.12 for females). Finally, lizard size (i.e., snout-vent length) did not have a significant effect on survivorship (χ21 = 2.94, P = 0.09).

Bottom Line: Color did not have a significant effect on survival, but we found several unexpected relationships including variation in predation over small spatial and temporal scales.In addition, we detected a marginally significant interaction between sex and color, suggesting selection for substrate matching may be stronger for males than females.We use our results as a case study to examine six major challenges frequently encountered in field-based studies of natural selection, and suggest that insight into the complexities of selection often results when experiments turn out differently than expected.

View Article: PubMed Central - PubMed

Affiliation: Department of Biological Sciences, University of Idaho, Moscow, Idaho, United States of America.

ABSTRACT
Determining the adaptive significance of phenotypic traits is key for understanding evolution and diversification in natural populations. However, evolutionary biologists have an incomplete understanding of how specific traits affect fitness in most populations. The White Sands system provides an opportunity to study the adaptive significance of traits in an experimental context. Blanched color evolved recently in three species of lizards inhabiting the gypsum dunes of White Sands and is likely an adaptation to avoid predation. To determine whether there is a relationship between color and susceptibility to predation in White Sands lizards, we conducted enclosure experiments, quantifying survivorship of Holbrookia maculate exhibiting substrate-matched and substrate-mismatched phenotypes. Lizards in our study experienced strong predation. Color did not have a significant effect on survival, but we found several unexpected relationships including variation in predation over small spatial and temporal scales. In addition, we detected a marginally significant interaction between sex and color, suggesting selection for substrate matching may be stronger for males than females. We use our results as a case study to examine six major challenges frequently encountered in field-based studies of natural selection, and suggest that insight into the complexities of selection often results when experiments turn out differently than expected.

Show MeSH