Growth Rate Variation Among Passerine Species in Tropical and Temperate Sites: an Antagonistic Interaction Between Parental Food Provisioning and Nest Predation Risk

ORIGINAL ARTICLE

doi:10.1111/j.1558-5646.2011.01227.x

GROWTH RATE VARIATION AMONG PASSERINE SPECIES IN TROPICAL AND TEMPERATE SITES: AN ANTAGONISTIC INTERACTION BETWEEN PARENTAL FOOD PROVISIONING AND NEST PREDATION RISK

Thomas E. Martin,1,2 Penn Lloyd,3,4,5 Carlos Bosque,6,7 Daniel C. Barton,3,8 Atilio L. Biancucci,3,9,10 Yi-Ru Cheng,3,11,12 and Riccardo Ton3,13 1U. S. Geological Survey Montana Cooperative Wildlife Research Unit, University of Montana, Montana 59812 2E-mail: [email protected] 3Montana Cooperative Wildlife Research Unit, University of Montana, Missoula, Montana 59812 4Percy FitzPatrick Institute of African Ornithology, DST/NRF Centre of Excellence, University of Cape Town, Rondebosch 7701, South Africa 5E-mail: [email protected] 6Departamento Biologıa´ Organismos, Universidad Simon Bolivar, Caracas, Venezuela 7E-mail: [email protected] 8E-mail: [email protected] 9E-mail: [email protected] 11E-mail: [email protected] 13E-mail: [email protected]

Received September 16, 2010 Accepted January 3, 2011

Causes of interspeciﬁc variation in growth rates within and among geographic regions remain poorly understood. Passerine birds represent an intriguing case because differing theories yield the possibility of an antagonistic interaction between nest predation risk and food delivery rates on evolution of growth rates. We test this possibility among 64 Passerine species studied on three continents, including tropical and north and south temperate latitudes. Growth rates increased strongly with nestling predation rates within, but not between, sites. The importance of nest predation was further emphasized by revealing hidden allometric scaling effects. Nestling predation risk also was associated with reduced total feeding rates and per-nestling feeding rates within each site. Consequently, faster growth rates were associated with decreased per-nestling food delivery rates across species, both within and among regions. These relationships suggest that Passerines can evolve growth strategies in response to predation risk whereby food resources are not the primary limit on growth rate differences among species. In contrast, reaction norms of growth

10Present address: P.O. Box 1927, Normal, Alabama 35762. 12Present address: Endemic Species Research Institute, Ming-shen East Road, Chichi Township, Nantou County, Taiwan.

C 2011 The Author(s). Evolution C 2011 The Society for the Study of Evolution. 1607 Evolution 65-6: 1607–1622 THOMAS E. MARTIN ET AL.

rate relative to brood size suggest that food may limit growth rates within species in temperate, but not tropical, regions. Results here provide new insight into evolution of growth strategies relative to predation risk and food within and among species.

KEY WORDS: Food limitation, growth rates, life history, nest predation, parental care, passerines, provisioning rate.

Growth rates of offspring vary extensively among species of all Martin 2004). The same relationships are expected across species taxa (Case 1978; Roff 1992; Arendt 1997; Dmitriew 2011). Inter- where evolution of slower growth can adjust food demands to specific variation in growth and development rates is even greater supply rates (Lack 1968; Martin 1987), but tests across species across geographic regions, such as tropical versus temperate (e.g., are lacking. Ricklefs 1976; Case 1978; Martin and Schwabl 2008; Cox and Food delivery rates may not be influenced by food availabil- Martin 2009). Causes of variation in growth rates of offspring ity alone. Increased nest predation risk can cause fewer feeding are important to understand because growth rates can influence trips by parents to reduce the risk of predators discovering nests offspring phenotypes and quality, thereby creating major fitness (Fig. 1B; Skutch 1949; Martin et al. 2000a,b; Eggers et al. 2005; consequences (Linden´ et al. 1992; Roff 1992; Arendt 1997). Yet, Fontaine and Martin 2006; Massaro et al. 2008). Greater pre- our understanding of potential causes of growth rate variation dation risk, therefore, can cause a proximate slowing of growth among species with diverse life histories within and among geo- via reduced food delivery within bird species (Scheuerlein and graphic regions remains weak. Gwinner 2006; Thomson et al. 2006), as commonly seen in other Two ecological factors (food limitation and juvenile pre- taxa (reviewed in Dmitriew 2011). Similarly, species with re- dation risk) are thought to play key roles in growth rate vari- duced feeding rates from higher predation risk may evolve slower ation within constraints imposed by allometric scaling across growth to adjust food demands to supply rates. The slower growth taxa (Case 1978; Arendt 1997; Dmitriew 2011). Food Limita- observed for many tropical species (i.e., Ricklefs 1976; Cox and tion: Greater food limitation can serve as a proximate cause Martin 2009) might reflect such effects given commonly higher of slower growth within species (Lack 1968; Crossner 1977; predation risk in many tropical sites and expected reductions in Martin 1987; Richner et al. 1989; Arendt 1997; Naef-Daenzer feeding rates (i.e., Skutch 1949; Martin 1996). Reductions in feed- and Keller 1999; McAdam and Boutin 2003). Greater food limi- ing rates from increased risk of predation, thus, may cause both tation also may favor evolution of slower growth among species, proximate and evolutionary slowing of growth rates (Fig. 1A,B). although such possibilities are relatively uninvestigated (Case Increased nest predation risk, however, is expected to favor 1978; Arendt 1997). Juvenile predation: Greater risk of juvenile evolution of faster, rather than slower, growth to minimize expo- predation can favor evolution of faster growth to minimize expo- sure time to predators (Fig. 1C; Case 1978; Bosque and Bosque sure time to predators (Williams 1966; Lack 1968; Case 1978; 1995; Remesˇ and Martin 2002). Bird species differ in nest pre- Bosque and Bosque 1995; Martin 1995; Arendt 1997; Remesandˇ dation risk based on their nest sites (Martin 1995; Fontaine et al. Martin 2002). Yet, increased predation risk in many systems 2007), and exhibit faster growth and shorter nestling periods when causes reduced feeding activity, which can slow growth rate via risk of nest predation is higher (Bosque and Bosque 1995; Remesˇ food limitation (Van Buskirk 2000; Altwegg 2002; Benard 2004; and Martin 2002; Ferretti et al. 2005). Food delivery could fa- see review in Dmitriew 2011). Thus, the relative importance and cilitate faster growth with increased predation risk if per-nestling interactions of offspring predation and food limitation may be feeding rates increased with nest predation risk (Fig. 1D) through critical to evolution of growth strategies but, as described below, mechanisms such as decreased brood size (number of nestlings) may include alternatives that have not been considered. (Slagsvold 1982; Martin 1995; Martin et al. 2000a; Doligez and Altricial birds provide an interesting group to study food lim- Clobert 2003; Eggers et al. 2006). However, no study has ex- itation and offspring predation because their relative influences amined per-nestling feeding rates relative to nest predation risk have been widely debated (reviewed in Martin 1987, 1992, 1996; across species. Lima 2009; Martin and Briskie 2009). Food limitation is com- An antagonistic interaction between food delivery and nest monly invoked as the dominant influence in birds (Lack 1968; predation (Fig. 1E) could arise if growth rates increase (Fig. 1C) Martin 1987) with expectations that growth rates increase with whereas per-nestling feeding rates decrease (Fig. 1F; e.g., Skutch the rates that parents deliver food per offspring (Fig. 1A). Greater 1949; Sargent 1993) with increasing predation risk. Independence food abundance can yield increased food delivery rates by parents of growth rates and food delivery is possible through evolution and faster growth rates within bird species at a proximate level of strategies that shift allocation of resources between growth (Naef-Daenzer and Keller 1999; Tremblay et al. 2003; Lloyd and rate and development of internal systems (Ricklefs 1968, 1993;

1608 EVOLUTION JUNE 2011 GROWTH RATE VARIATION

Figure 1. Three alternative hypotheses for evolutionary inﬂuences of nest predation and food limitation on growth rates, holding any mass effects constant. (A) If food limitation exerts selection on evolution of growth rates, growth rate should increase with per-nestling feeding rate (Naef-Daenzer and Keller 1999; Tremblay et al. 2003; Lloyd and Martin 2004). (B) Feeding rates may be inﬂuenced by nest predation risk or food availability. If nest predation exerts selection on evolution of feeding rates, then total feeding rates (total number of feeding-trips by both parents h−1) should decrease with increased predation risk (Skutch 1949; Martin et al. 2000a,b; Fontaine and Martin 2006). An additive effect of food availability will cause deviations in total feeding rates for a given nest predation rate (the gray arrows). A dominant role of food availability could eliminate a relationship with nest predation. (C) If nest predation exerts selection on evolution of growth rates, species with higher nest predation risk are expected to evolve increased growth rates (Case 1978; Arendt 1997; Remesˇ and Martin 2002). (D) Per-nestling provisioning may increase with nest predation risk if brood size is also reduced. This result can allow increased growth rates with greater predation risk through reduced food limitation (see A) as a facilitating mechanism. An additive effect of food availability should cause deviations in per-nestling feeding rates (gray arrows) for a given predation rate. (E) Growth rate may vary antagonistically with per-nestling feeding rate if nest predation favors increased growth rates but reduced per-nestling feeding rates (see F). (F) Per-nestling feeding rates might decrease with increased nest predation risk, similar to total feeding rates (Skutch 1949; Martin 1996).

Arendt 1997). Indeed, species can evolve strategies of resource sizes (e.g., Shkedy and Safriel 1992). Finally, growth rates may allocation that yield faster growth rates among species or popula- increase with brood size (Fig. 2) if: (1) food is not limiting, (2) par- tions at higher risk of predation even with constant food (Arendt ents reduce feeding in response to reduced begging cues (Hinde 1997; McPeek 2004; Walsh and Reznick 2008). Yet, such possi- et al. 2010), (3) poorer parents raise fewer young and are also bilities are untested across broad suites of species. poorer at provisioning them, or (4) parents reduce per-nestling The dominance of food limitation in theories related to birds feeding rates with reductions in brood size (e.g., Ghalambor and (Lack 1968; Martin 1987) has been based primarily on studies Martin 2000; Chalfoun and Martin 2010). of proximate relationships within-species rather than evolved dif- We examine the alternative hypotheses and a priori predic- ferences among species. Proximate responses may differ from tions, outlined in Figure 1, based on our studies of growth rates in evolutionary responses to an environmental factor (Martin 2004). 64 species of Passerine birds at sites on three continents. More- Reaction norms of growth rates relative to brood size variation over, we examine interspecific variation in the intraspecific reac- within species (Fig. 2) provide an opportunity to contrast the in- tion norms of growth rates relative to brood size (see Fig. 2) to fluence of food limitation on variation in growth rates within- contrast possible roles of food limitation constraints on growth versus among-species. A decrease in growth rate with increasing rates within versus among species. brood size (Fig. 2) indicates increasing food limitation (Cross- ner 1977; Martin 1987; Moreno et al. 1998; Kunz and Ekman Methods 2000; Tremblay et al. 2003). In contrast, if parents adjust brood STUDY SYSTEM sizes to their ability to provision offspring (Pettifor et al. 2001), We studied growth rates of 64 Passerine species (Appendix S1) at growth rates and food limitation should be similar among brood sites on three continents. In particular, 18 species were studied in

EVOLUTION JUNE 2011 1609 THOMAS E. MARTIN ET AL.

knew the exact days of hatch and fledge to provide particularly precise estimates of nestling periods. Nest predation was assumed when all nestlings disappeared more than two days prior to average fledging age, and parents could not be found in the area feeding fledglings. Nest predation was calculated as daily predation rate (Mayfield 1975; Hensler and Nichols 1981) during the nestling period. We videotaped nests for 6–8 h starting within 30 min of sun- rise to quantify parental feeding rates as total feeding-trips h−1 (Martin et al. 2000a). Numbers of feeding-trips to the nest have been found to be good predictors of food delivery and nestling growth (Lyon et al. 1987; Nolan et al. 2001; MacColl and Hatch- well 2003). Per-nestling feeding rates (i.e., parental provisioning per offspring) were measured as feeding-trips nestling−1 h−1 by Figure 2. Theoretical predictions of intraspeciﬁc reaction norms dividing the total feeding rate at a nest by the number of nestlings of growth rates relative to brood size based on differing levels of in that nest. To avoid behavioral disturbance, video cameras were food limitation. An absence of food limitation, or equivalent lev- left unattended 4–10 m from the nest (using optical zooms up to els of food limitation under the individual optimization hypoth- 30×). In addition, the camera was often camouflaged with a cover esis (i.e., Pettifor et al. 2001), should yield similar growth rates and the tripod was hidden by natural vegetation. We videotaped among brood sizes (Tremblay et al. 2003) and a slope approximat- ing 0. Also, nest predation may favor maximum growth rates such parental feeding rates during the nestling stage, at days 2–3, on that growth rate does not change with brood size (Shkedy and the day that pin feathers broke their sheaths, and two to three days Safriel 1992). Alternatively, food limitation causes growth rate to prior to the average fledging date (Martin et al. 2000a). However, increase (slope < 0) with reduced brood size (Lack 1968; Martin to standardize developmental stage, we only use data from nests 1987; Moreno et al. 1998; Tremblay et al. 2003). Finally, growth sampled when primary pin feathers broke their sheaths. Pin-break > rates may increase with brood size (slope 0) if food is not limit- varies among nests within species in a two- to three-day window, ing, or parents reduce feeding in response to reduced begging cues but analysis of covariance (ANCOVA) tests indicated that vari- (Hinde et al. 2010), or parents reduce per-nestling feeding rates ation in age within pin-break did not contribute to variation in with reductions in brood size as might be expected in long-lived > species (e.g., Ghalambor and Martin 2000; Fontaine and Martin feeding rate at any site (P 0.4 in all three sites). 2006; Chalfoun and Martin 2010). We weighed nestlings using ACCULAB (Elk Grove, IL) portable electronic scales with an accuracy of ±0.001 g (Fierro- high-elevation (2300 m elevation) mixed forest in Arizona (34oN Calderon´ and Martin 2007; Cox and Martin 2009). We weighed latitude), 29 species in cloud forest (1350–2000 m) in the northern nestlings every other day beginning on day of hatch or within Andes of Venezuela (9oN), and 17 species in coastal dwarf shrub two days after hatch. We estimated growth rates using the logis- land in southern South Africa (34oS) (see Fierro-Calderon´ and tic growth curve because growth is typically S-shaped and this Martin 2007; Martin 2007; Martin et al. 2007 for descriptions of approach only produces three parameters that are readily bio- study sites). logically interpretable based on the equation: W(t) = A/(1 + e ∗ (−K (t−ti))), where W(t) denotes body mass of a nestling at time

FIELD SAMPLING t, A is the asymptotic mass that nestlings approach, ti is the in- We searched study sites for nests during entire breeding seasons flection point of the curve, and K is a constant scaling rate of in 1988–2009 in Arizona, 2000–2004 in South Africa, and 2002– growth (Ricklefs 1968; Remesˇ and Martin 2002). K, the growth 2008 in Venezuela. For species examined here, large numbers of rate constant, is a standardized measure of growth rate that is inde- nests were located and monitored in Arizona (10,157 nests), South pendent of absolute time and size and, thus, is particularly useful Africa (4385 nests), and Venezuela (3289 nests) following long- for comparative studies (Ricklefs 1968; Starck and Ricklefs 1998; term protocols (Martin 2002, 2007; Martin et al. 2000a,b, 2007). Remesˇ and Martin 2002). Moreover, we tested K against length Nests were checked every two to four days to determine status of the nestling period as an independent measure of relative rate and predation events, but were checked daily or twice daily near of development to test the validity of K as a measure of relative stage-changing events such as completion of laying, hatching, and differences in growth rates. We also obtained mass of adults by fledging to determine period lengths. We quantified the nestling capturing them in mist-nets. Captured adults were weighed using period as the number of days from when the last egg hatched and pesola scales or ACCULAB (Elk Grove, IL) portable electronic the last nestling fledged and we did this only for nests where we scales to obtain adult mass in each site.

1610 EVOLUTION JUNE 2011 GROWTH RATE VARIATION

ANALYSES transformed dummy variables to test site as a main effect and the We used ANCOVA to explicitly test hypotheses of variation in other variables as covariates. The cumulative change in Sums of growth rates and parental feeding rates, with site as a random Squares when the two site dummy variables were entered as a factor, and nest predation and log-transformed body mass as group was then examined to determine statistical significance of possible covariates. Two species (Mionectes olivaceus, Masius site. The phylogenetic hypothesis was based on the most recent chrysopterus) had a lek-breeding social system, where females information available for relevant taxa (e.g., Sibley and Ahlquist provided care alone, whereas males helped to feed in all other 1990; Burns et al. 2002; Yuri and Mindell 2002; Klicka et al. species. Given this solitary care, we might expect feeding rates to 2005; Voelker et al. 2007; Tello et al. 2009; also summarized in be lower and so we included a dummy variable for female-only Martin et al. 2006) and branch lengths were set to equal lengths care in models. Even though only two species had female-only because standardized branch lengths were not available for all care, they were sufficiently different that this was consistently a species. significant variable in the model (see results) and altered estimates of slopes in Venezuela and therefore estimates of interactions. We included interactions with site, but subsequently removed interac- Results tions and do not report them when they were clearly not significant NESTLING PERIOD TEST (i.e., P > 0.10). We did not remove main factors when they were The duration of the nestling period (days from hatch to fledge) is not significant, because the main factors reflect explicit hypothe- a result of nestling growth rates interacting with selection pres- ses being tested. Where interactions were significant (P < 0.05) sures such as nest predation favoring fledging at differing stages or marginally significant (i.e., 0.05 < P < 0.10), we further ana- of development (Bosque and Bosque 1995; Martin 1995; Remesˇ lyzed each site separately. We estimated effect size in ANCOVAs and Martin 2002; Roff et al. 2005). We tested the expected cor- as partial eta-squared that represented the proportion of varia- relation between nestling period and growth rate to use an in- tion explained by a variable (SPSS 2009). We also examined the dependently measured trait (nestling period) to verify biological possible influences of covariances among variables while testing validity of growth rate estimates. Growth rates were inversely cor- predicted relationships using structural equation modeling, but related with nestling periods within and among sites, as expected results did not differ qualitatively from ANCOVA and, therefore, (Table 1; Fig. 3). Given the significant interaction due to the dif- are presented in Appendix S2. fering slope in South Africa, we tested each site separately and We controlled for possible phylogenetic effects using inde- found strong correlations in each site (Fig. 3). The two species pendent contrasts (IC) (Felsenstein 1985) from CAIC (Purvis and with female-only care in Venezuela had slow growth but their Rambaut 1995), and a regression approach to ANCOVA (Grafen long nestling periods fit the relationship such that this was not

1989, 1992; Martin 1995, 2002). Two dummy variables were cre- a significant factor in Venezuela (F1,24 = 0.1, P = 0.76). Thus, ated for sites and one dummy variable for female-only care and the independently measured nestling periods supported growth IC were calculated for these dummy variables along with the life- rate estimates (Fig. 3), and we focused remaining analyses only history variables for all species. Multiple regression models were on growth rate given the high redundancy between these two used on the IC to conduct ANCOVAs using the phylogenetically variables.

Table 1. Results of ANCOVA to test covariance of growth rates with length of the nestling period, while controlling for any effects of mass and species with female-only care based on raw data and independent contrasts from Felsenstein (1985).

Raw data Independent contrasts Variables df F-value P-value Effect size1 df F-value P-value

Nestling period 1 320.9 <0.001 0.86 1 207.4 <0.001 Mass 1 0.3 0.60 0.01 1 0.03 0.87 Female-only care 1 0.2 0.67 0.01 1 0.04 0.84 Site 2 13.8 <0.001 0.35 2 9.4 <0.001 Site×Nestling period 2 10.7 <0.001 0.30 2 8.3 0.001 Error 51 51

1Effect size is partial Eta-squared, or the proportion of variation explained (SPSS 2009).

EVOLUTION JUNE 2011 1611 THOMAS E. MARTIN ET AL.

sidered alone did not explain growth rates (Fig. 4) among the groups of species studied here, although growth rates differed among sites and were slower in the two species with female-only care (Table 2A). Species in north temperate Arizona had faster growth than tropical Venezuela (LSD, P = 0.001) and marginally faster than south temperate South Africa (LSD, P = 0.099), whereas South Africa and Venezuela did not quite differ (LSD, P = 0.11). Behavior also can show scaling (Dial et al. 2008), so we examined total feeding activity (feeding-trips h−1) and per- nestling feeding rates (feeding-trips nestling−1 h−1) relative to log-transformed mass. In general, we found that larger species had lower feeding activity (Table 2B; Fig. 5A). However, examination of each site separately showed that Venezuela was the only significant individual site (Fig. 5A). We also found a strong relationship between mass and per- K Figure 3. Growth rate ( ) relative to nestling period. Duration of nestling feeding rates (Table 2C; Fig. 5B). The interaction was not the nestling period (number of days from hatch to fledge) is an significant, but visual inspection (Fig. 5B) suggested that slopes independently measured trait that should reflect growth rate (K) differed, so we analyzed each site separately. Again, analyses of and the relationships are strong in all three sites. The two open symbols reflect two species in Venezuela with lek-breeding and the individual sites showed that only Venezuela had a significant female-only care. relationship (Fig. 5B).

NEST PREDATION EFFECTS ON GROWTH RATE AND ALLOMETRIC SCALING FEEDING RATES We tested growth rates for possible allometric scaling constraints We tested the predictions that nest predation risk can favor evo- across the three sites, where body mass distributions were rea- lution of differences among species in their growth rates, total = = sonably similar (Fig. 4; F2,59 2.4, P 0.10). The growth rate feeding activity, and per-nestling feeding rates (Fig. 1A,B,D,F). constant (K) is expected to be negatively related to adult mass, re- Nest predation rates explained a large percentage of the variation flecting slower growth among larger species (Case 1978; Starck in growth rates across species in all three sites but slopes differed and Ricklefs 1998). However, log-transformed body mass con- among sites (Table 3A; Fig. 6A). Given the interaction with site in the raw data, we also tested the effect of nest predation for each site separately and found that relationships were strong in each site (Fig. 6A). Results also showed that growth rates were low for the species with female-only care (Table 3A; Fig. 6A). Finally, after accounting for effects of nest predation, a marginally significant negative allometric relationship was expressed by growth rate (Table 3A). Total feeding rates strongly decreased with nest predation risk in all three sites (Fig. 6B), and with body mass (Fig. 5A) and female-only care (Fig. 6B), and differed among sites (Table 3B). Given the significant interaction with site based on raw data, we also tested the effect of nest predation for each site separately while controlling possible effects of mass through partial correlation analysis and found that predation remained a significant predictor of feeding rate in all three sites (Fig. 6B). Mass was a marginally significant covariate in Arizona Figure 4. Growth rate (K) relative to adult body mass. The typical r =− P = allometric expectation of decreased growth rate in larger species (mass: p 0.46, 0.061), but nestling predation remained is not met in this set of species either within any site or across strong (Fig. 6B). Arizona also had one strong and high outlier sites. The two open symbols reﬂect two species in Venezuela with (Fig. 6B) that was a flycatcher (Empidonax occidentalis). Unusu- lek-breeding and female-only care. ally high feeding activity of flycatchers related to their foraging

1612 EVOLUTION JUNE 2011 GROWTH RATE VARIATION

− Table 2. Results of ANCOVA to test allometric scaling relationships for growth rate, total feeding rate (feeding-trips h 1), and per- nestling investment (feeding-trips nestling−1 h−1) based on raw data and independent contrasts from Felsenstein (1985).

Raw data Independent contrasts Variables df F-value P-value Effect size df F-value P-value

(A) Growth rate Mass 1 0.1 0.74 0.00 1 2.1 0.16 Female-only care 1 5.5 0.022 0.09 1 3.5 0.067 Site 2 5.7 0.005 0.18 2 6.2 0.004 Error 54 54 (B) Total feeding rate (feeding-trips h−1) Mass 1 9.9 0.003 0.15 1 4.4 0.040 Female-only care 1 1.8 0.18 0.03 1 3.5 0.067 Site 2 2.6 0.085 0.08 2 7.3 0.001 Error 57 57 (C) Per-nestling investment (feeding-trips nestling−1 h−1) Mass 1 9.0 0.004 0.14 1 13.4 0.001 Female-only care 1 3.2 0.077 0.05 1 2.7 0.107 Site 2 2.7 0.076 0.09 2 1.6 0.20 Error 57 57

− Figure 5. Feeding activity relative to body mass. Relationships of (A) total feeding rate (feeding-trips h 1) and (B) per-nestling feeding rates (feeding-trips nestling−1 h−1) relative to log-transformed mass. The two open symbols reﬂect two species in Venezuela with lek- breeding and female-only care. The partial correlations exhibited for Venezuela control for the two species with female-only care which had (A) lower total feeding rates (F1,24 = 3.8, P = 0.065) and (B) lower per-nestling feeding rates (F1,24 = 4.8, P = 0.038) for their body masses.

EVOLUTION JUNE 2011 1613 THOMAS E. MARTIN ET AL.

− Table 3. Results of ANCOVA to explain variation in growth rates, total feeding rates (feeding-trips h 1), and per-nestling investment (feeding-trips nestling−1 h−1) by nest predation while controlling for any effects of mass and female-only care based on raw data and independent contrasts from Felsenstein (1985).

Raw data Independent contrasts Variables df F-value P-value Effect size df F-value P-value

(A) Growth rate Nest predation rate 1 52.7 <0.001 0.52 1 26.4 <0.001 Mass 1 3.4 0.073 0.06 1 3.7 0.059 Female-only care 1 7.6 0.008 0.13 1 3.0 0.089 Site 2 1.8 0.17 0.07 2 13.5 <0.001 Site×predation 2 3.3 0.047 0.12 ns Error 49 51 (B) Total feeding rate (feeding-trips h−1) Nest predation rate 1 59.0 <0.001 0.53 1 29.7 <0.001 Mass 1 4.0 0.051 0.07 1 2.0 0.17 Female-only care 1 4.7 0.034 0.08 1 5.2 0.027 Site 2 3.4 0.042 0.11 2 3.5 0.037 Site×predation 2 3.5 0.038 0.12 ns Error 52 54 (C) Per-nestling investment (feeding-trips nestling−1 h−1) Nest predation rate 1 23.3 <0.001 0.30 1 11.5 0.001 Mass 1 2.3 0.13 0.04 1 1.9 0.17 Female-only care 1 6.4 0.015 0.11 1 17.4 <0.001 Site 2 10.1 <0.001 0.27 2 5.6 0.006 Error 54 54

behavior has been recognized previously (Martin et al. 2000a). and both mass and female-only care also explained variation

Inclusion of a flycatcher dummy variable (F1,14 = 33.3, P < (Table 4A; Fig. 7). The model also indicated that growth rates

0.001) and adult mass (F1,14 = 6.4, P = 0.024) caused the nega- scaled negatively with log-transformed body mass (b ± SE = tive relationship between total feeding activity and nest predation −0.028 ± 0.012). Given the significant interaction, we tested to become very strong (nestling predation: rp =−0.94, P < each site separately. The negative relationships of growth rate 0.001; Fig. 6B). Mass was not a significant covariate in South with per-nestling feeding rates were significant in all three sites Africa or Venezuela, but the two species with female-only care tested separately (Fig. 7). in Venezuela had low feeding rates for their nest predation rates Nestling predation was included (Table 4B) to test the pos- (Fig. 6B). sible interaction and dominance of predation versus feeding rates Per-nestling feeding rates decreased with increased nest pre- (i.e., Fig. 1A vs. E). Nestling predation was the strongest predictor dation risk (Fig. 6C), and was lower in species with female- of growth rates, but per-nestling feeding rates, female-only care only care, whereas mass and interactions were not significant and body mass all contributed to explaining variance (Table 4B). (Table 3C; Fig. 6C). As a result, relationships were strong in each When nest predation was included in the model, the explana- site considered separately (Fig. 6C). tory power of per-nestling feeding rates became much weaker in the raw data and disappeared in the phylogenetically corrected IC GROWTH RATE RELATIVE TO PER-NESTLING (Table 4B). The relationship of growth rate with per-nestling feed- FEEDING RATES ing rates in the raw data remained negative (b ± SE =−0.010 ± Growth rates should either increase or decrease with per-nestling 0.004) even after accounting for the other effects in the model feeding rates to reflect whether food limitation or nest predation, (i.e., Table 4B). These analyses also showed that once the strong respectively, is dominant in driving growth rate evolution (Fig. 1A effects of nest predation were included, growth rates showed neg- vs. E). Growth rate decreased with increased per-nestling feed- ative allometric scaling with log-transformed mass (b ± SE = ing rates in all three regions, but slopes differed among regions, −0.024 ± 0.010).

1614 EVOLUTION JUNE 2011 GROWTH RATE VARIATION

− Figure 6. Nest predation effects. Relationships of (A) growth rate constant (K), (B) total feeding rate (feeding-trips h 1), and (C) per- nestling feeding rates (feeding-trips nestling−1 h−1) with nest predation rates during the nestling period. The two open symbols reﬂect two species in Venezuela with lek-breeding and female-only care. The partial correlations exhibited for Venezuela control for the two species with female-only care that have (A) lower growth rates (F1,22 = 4.6, P = 0.043), (B) lower total feeding rates (F1,22 = 9.2, P = 0.006), and (C) lower per-nestling feeding rates (F1,22 = 6.2, P = 0.021) for their nest predation rates. The partial correlation for Arizona in B controls for the effects of log-transformed mass on feeding rate (see results).

REACTION NORMS OF GROWTH RATES WITH BROOD R2 = 0.90; Fig. 8). The highly significant interaction shows that SIZE species have differing reaction norm responses in growth rate Growth rates were calculated for every brood size (number of to brood sizes. We then examined whether these reaction norm nestlings in the brood) possible. This yielded 11 species in slopes differed among the sites. Arizona, 11 in South Africa, and 19 species in Venezuela with Only one of the 11 species in Arizona (i.e., Pipilo chlorurus) growth rates (K) for more than one brood size (Fig. 8). We cal- had a positive slope (Fig. 8A) and the slopes averaged (mean ± culated the slopes of the change in growth rate with brood size SE =−0.020 ± 0.006) significantly less than 0 (t10 =−3.47, for each species as the interaction between species and brood size P = 0.006). South Africa had eight of 11 species with negative using ANCOVA, and including site as a factor and brood size as slopes (Fig. 8B), yielding an average negative slope (mean ± a covariate (site: F2,59 = 25.1, P < 0.001; brood size: F1,59 = SE =−0.010 ± 0.005) that was marginally less than 0 (t10 =

0.5, P = 0.48; species × brood size: F41,59 = 9.0, P < 0.001; −1.85, P = 0.095). Venezuela, in contrast, had positive slopes in

EVOLUTION JUNE 2011 1615 THOMAS E. MARTIN ET AL.

− Table 4. Results of ANCOVA to explain variation in growth rates when including per-nestling feeding rate (feeding-trips nestling 1 h−1) when nestling predation (A) is not or (B) is included in the model based on raw data and independent contrasts.

Raw data Independent contrasts Variables df F-value P-value Effect size df F-value P-value

(A) Nestling predation not included Per-nestling feeding rate 1 28.8 <0.001 0.39 1 9.4 0.004 Female-only care 1 17.6 <0.001 0.29 1 10.4 0.002 Mass 1 5.7 0.021 0.09 1 2.9 0.092 Site 2 4.8 0.012 0.16 2 3.9 0.026 Site×feeding rate 2 4.2 0.021 0.16 2 3.4 0.043 Error 49 49 (B) Nestling predation included Per-nestling feeding rate 1 7.2 0.011 0.14 1 1.5 0.23 Nest predation rate 1 25.9 <0.001 0.36 1 17.3 0.004 Female-only care 1 11.9 0.001 0.21 1 3.8 0.059 Mass 1 6.3 0.016 0.12 1 3.9 0.053 Site 2 1.2 0.30 0.05 2 10.9 <0.001 Site×predation 2 4.0 0.026 0.15 ns Error 46 48

13 of the 19 species (Fig. 8C), yielding an average positive slope (LSD, P = 0.001) and South Africa (LSD, P = 0.012) differing

(mean ± SE = 0.025 ± 0.010) that was greater than 0 (t18 = 2.42, from Venezuela, but Arizona and South Africa not differing from P = 0.026). As a result, the three sites differed in their average each other (LSD, P = 0.49). reaction norm slopes (F2,38 = 7.1, P = 0.002), with Arizona

Discussion The importance of nest predation to evolution of growth rate variation has been debated, with both positive and negative evidence (e.g., Lack 1968; Ricklefs 1968, 1976; Bosque and Bosque 1995; Starck and Ricklefs 1998; Remesˇ and Martin 2002). The tests across diverse species and sites here suggest that nest predation is a dominant influence on evolutionary divergence in growth rates among species. Nest predation was a strong predictor and explained a large percent of the variation in growth rate among species (Figs. 1C and 6A). Growth rates were faster in species that had higher predation risk but lower per-nestling feeding rates (i.e., food delivery per nestling) (Fig. 7), which also argues for a dominant role of nest predation (Fig. 1E). Numbers of provisioning trips have been found to be a good reflection of the total mass of food delivered to offspring and related to nestling growth and survival (Lyon et al. 1987; Nolan et al. 2001; MacColl and Hatchwell 2003). Moreover, the rate Figure 7. Growth rate relative to per-nestling feeding rates of provisioning at pin-break is highly correlated with the rate of (feeding-trips nestling−1 h−1). Growth rates increase with decreas- provisioning early (i.e., 1–3 days after hatching) in the nestling ing food delivery rate per nestling in all three sites. The two open = < = symbols reﬂect two species in Venezuela with lek-breeding and period (r 0.833, P 0.001, n 58 spp) indicating that species female-only care. The partial correlation exhibited for Venezuela remain reasonably consistent in feeding rate differences across the controls for the two species with female-only care, which had nestling period. Yet, species with a lower rate of provisioning trips lower growth rates (F1,22 = 11.5, P = 0.003) for their feeding rates. associated with increased predation risk may compensate to some

1616 EVOLUTION JUNE 2011 GROWTH RATE VARIATION

Figure 8. Reaction norms of growth rate relative to brood size within species for differing species in our three study sites: (A) Arizona, (B) South Africa, and (C) Venezuela. extent by increasing the amount (load) of food per trip (Martin The increase in growth rates despite lower food delivery et al. 2000a). Still, increases in food load do not fully compensate (Fig. 7) suggests an antagonistic interaction between nest preda- for decreased numbers of provisioning trips such that total food tion and food limitation on growth rates (Fig. 1E,F) that is particu- delivered declines with declining numbers of trips among species larly interesting. The results here contrast with those of many stud- (Martin et al. 2000a, unpubl. data). Even if food loads fully com- ies in other taxa that also found decreased feeding activity with pensated for decreased feeding trips, the resulting similar food increased predation risk, but found a correlated decrease, rather delivery rates cannot explain the differing growth rates under than increase, in growth rates (see review in Dmitriew 2011; also the food limitation hypothesis. Finally, food quality might differ bird studies by Scheuerlein and Gwinner 2006; Thomson et al. among species, but we are comparing species within a single habi- 2006). Yet, such effects are proximate responses observed within tat in each site and many feed the same foods (e.g., lepidopteran species that contrast with the evolved cross-species results here. larvae) to their offspring with no evidence of a systematic increase The increase in growth rates with lower food delivery observed in food quality with declining feeding rate. Thus, the increase in here in Passerine species with high nest predation risk may sug- growth rate with reduced per-nestling provisioning rate (Fig. 7) gest evolution of growth strategies that differentially prioritize appears to conflict with the food limitation hypothesis. resource allocation to growth rate based on juvenile mortality risk

EVOLUTION JUNE 2011 1617 THOMAS E. MARTIN ET AL.

(see Arendt 1997; McPeek 2004; Walsh and Reznick 2008). For as the other two sites (Fig. 6A) and food delivery rates do not example, McPeek (2004) studied two damselflies and similarly seem to explain geographic variation. Thus, nest predation plays found faster growth despite the same food acquisition rate in one a strong role within regions, but is less important in explaining species with greater predation risk. The physiological basis of differences among regions where other selection pressures (e.g., such differences in growth rates are unknown, but investment in longevity) may become more important. resources for internal phenotypic components that enhance off- The generally negative reaction norm slopes of species in spring quality are known to create trade-offs with growth rate north (Arizona—Fig. 8A) and south (South Africa—Fig. 8B) (e.g., Arendt 1997, 2003; Badyaev and Martin 2000; Billerbeck temperate sites suggest food limitation within temperate species, et al. 2001; Brommer 2003; Soler et al. 2003; Ardia 2006). In- although the variation in slopes may reflect differing levels of creased nest predation risk may shift growth strategies toward food limitation among species. Increases in number of nestlings increased prioritization of growth rate over other phenotypic at- (i.e., brood size) were associated with decreased growth rates tributes. Thus, faster growth may incur phenotypic costs (e.g., suggesting that parents are working at the limits of parental ef- reduced investment in phenotypic quality—Arendt 1997, 2003; fort favored by natural selection and have increasing difficulty Starck and Ricklefs 1998; Arendt et al. 2001; Billerbeck et al. meeting resource needs of larger broods (also see Martin 1987; 2001; Dmitriew 2011), but these costs apparently can be over- Kunz and Ekman 2000). This result contrasts with the compar- shadowed by increased risk of predation, which can shift growth ative results showing that species with higher growth rates have strategies to prioritize faster growth rates even with reduced or lower food delivery rates (Fig. 7), favored by higher predation similar food delivery. risk. These contrasting results emphasize that ecological drivers Nest predation, on the other hand, was not a dominant in- of proximate variation within species are not necessarily the same fluence on geographic differences in growth rates; growth rates as the primary selection pressures driving evolution of variation differed among sites, with slowest growth in the tropics, for the among species (also Martin 2004). same nest predation rates (Fig. 6A; also see Ricklefs 1976; Cox Reaction norms in the tropical site (Fig. 8C) contrasted with and Martin 2009). Indeed, slower growth of tropical species in the two temperate sites by generally having positive, rather than other studies and locations are often associated with higher pre- negative, slopes. The slower growth in smaller brood sizes could dation risk compared with north temperate regions, further em- reflect poorer parents or suggest reduced per-nestling feeding rates phasizing the inability of predation to explain variation between in smaller broods. On the other hand, reduced growth in smaller regions (Ricklefs 1976; Martin 1996). The slower growth, con- broods could also reflect reduced sibling competition (Ricklefs trolling for nest predation, in the tropics could reflect an additive 1993; Lloyd and Martin 2003), or a reduced feeding response effect of greater food limitation. Yet, the per-nestling provision- by parents to reduced begging cues by fewer nestlings (Kolliker¨ ing rate is actually greater, rather than less, in the tropical species et al. 2000; Hinde et al. 2010). The contrasting results in the tropics (Fig. 6C; Table 3C; also see Martin et al. 2000a). Moreover, food compared with temperate sites clearly deserve further study. loads (volume of food feeding-trip−1) were at least as large as Skutch (1949) long ago proposed that nest predation risk for related species in the temperate sites (T.E. Martin, unpubl. could constrain evolution of parental feeding rates and that gen- data). On the other hand, food quality might be lower in the erally higher risk in the tropics favored reduced feeding that ex- tropics, but the positive reaction norms (faster growth rates in plained smaller clutch sizes there. This idea has generated much larger broods) in the tropics (Fig. 8C) suggest that food limita- discussion with many indirect tests over the years (Martin and tion does not explain intraspecific variation in growth rates in Briskie 2009). The first direct test of such possibilities between the tropics. Slower growth of tropical species, instead, might be the subtropics and temperate regions showed that nest predation explained by the greater longevity of many, but not all, tropical did indeed explain parental provisioning activity within but not birds (e.g., Johnston et al. 1997; Sandercock et al. 2000; Peach between regions (Martin et al. 2000a). We also show here that et al. 2001; Francis and Wells 2003; but see Karr et al. 1990; adult provisioning activity across species is indeed influenced by Blake and Loiselle 2008). High adult survival might favor growth predation risk, and such effects exist for a broad range of species strategies (i.e., slower growth) that allow physiological trade-offs and geographic sites (Fig. 6B). Moreover, the relationships were to increase offspring quality (Ricklefs 1993). Growth strategies much stronger with total provisioning activity (Fig. 6B; rp = clearly differ in tropical compared with north temperate species −0.69; P < 0.001) than with per-nestling feeding rates (Fig. 6C; given that growth rates are slower for the same nestling period rp =−0.38; P = 0.001). This difference follows directly from (Fig. 3), indicating that growth rate is not the only predictor of the hypothesis that total numbers of visits to the nest may in- when young birds leave the nest between regions. Although nest fluence risk of predation and be under selection (Skutch 1949; predation may not explain variation between regions, it is a strong Martin et al. 2000a). The hypothesis also depends on predators influence on variation in growth rates within the tropics, as well being diurnal and visually oriented. Indeed, video recordings of

1618 EVOLUTION JUNE 2011 GROWTH RATE VARIATION

predation events, observations in the field, and times of predation Fontaine and Martin 2006; Chalfoun and Martin 2010). Thus, in- events from data loggers all show that more than 90% of predation creased growth rates of species with higher predation risk cannot events occur during the day, especially morning, in both Arizona be explained by brood size effects on provisioning. and Venezuela (T. E. Martin, unpubl. data) and that visually ori- Finally, the diverse group of species studied here showed no ented predators dominate (also see Biancucci and Martin 2010). allometric scaling of growth rate when body mass was considered However, South Africa has a much greater incidence of noctur- alone (i.e., Fig. 4), as is the classical approach (e.g., Case 1978; nal snake predation, although it still has substantial diurnal bird Starck and Ricklefs 1998). Growth rates are expected to be slower and visually oriented mammal predation (T. E. Martin, unpubl. in larger species, and much evidence demonstrates such effects data). Nonvisually oriented snakes should exert less selection on (Case 1978; Starck and Ricklefs 1998). Yet, Starck and Ricklefs parental feeding activity and feeding rates are higher, on average, (1998) found that growth rates were not negatively related to mass controlling for nest predation (Fig. 6B—see results), and the re- in small (i.e., <100 g) Passerines, as we found here. However, we lationship shows much greater variance around the line than the found that growth rates scaled negatively with mass, as expected, other two sites. Indeed, the illustrated correlation (i.e., Fig. 6B) once we accounted for the effects of nest predation (Tables 3A, is strongly influenced by Anthoscopus minutus, which has a par- 4B). The obscuring of this allometric relationship when mass is ticularly safe nest, unusually large clutch size for this site, and a considered alone emphasizes the dominant role of nest predation high parental feeding activity (Fig. 6B—the left-most and highest in driving evolution of growth rates among small Passerines. point in the figure). If this species is removed, the correlation In summary, nest predation risk can play a strong role in the between feeding rate and nest predation drops substantially (r = evolution of phenotypic traits across diverse species (reviewed in −0.53, P = 0.032) and indicates that only 28% of the variance Lima 2009; Martin and Briskie 2009). Comparative field studies is explained. No single species yields such a strong drop in cor- allow examination of the different kinds of strategies that can relation strength for either of the other two sites, reflecting the evolve in response to differences in predation risk and these stronger overall relationships there. This weaker relationship in commonly contrast with ecological drivers of proximate variation South Africa where selection from visually oriented predators is within species. Our studies here suggest that evolutionary weaker reinforces the conclusion that visually oriented predators responses of growth rates and feeding rates to predation risk influence evolution of parental feeding strategies. yield a previously unrecognized antagonistic interaction that Although visual orientation of predators is important to evo- reinforces the dominant role of predation as a selective force on lution of parental feeding strategies, growth rates of offspring growth and provisioning strategies. Still, the slower growth in should be directly sensitive to mortality risk alone and visual tropical species independent of predation emphasizes that other orientation of predators is unimportant. Predation rates in South factors can dominate across geographic space and geographic Africa were as high as the other sites during the nestling period comparisons can improve understanding of the evolution of (Fig. 6) and were even higher than the other two sites when in- broad phenotypic variation. cluding the incubation period (Martin et al. 2007). Consequently, ACKNOWLEDGMENTS the relationship of South Africa growth rates with nest preda- We thank R. Duckworth, R. Hutto, H. Sofaer, and two anonymous re- tion rates was as strong as the other sites (Fig. 6A) even with a viewers for helpful comments on drafts of this manuscript. We gratefully weaker relationship between parental provisioning and predation acknowledge the help of the many field personnel that aided in collec- in South Africa. Such results reinforce the conclusion that feeding tion of these data over many years and sites. This work was supported rates have a weak influence on growth rates. by National Science Foundation grants (INT-9906030, DEB-9981527, DEB-0543178, DEB-0841764 to TEM) for studies in Arizona, South Nest predation can also favor a reduction in brood size Africa, and Venezuela, while the work in Arizona was also supported by (Slagsvold 1982; Martin 1995; Martin et al. 2000a) and if the de- the United States Geological Survey Climate Change Research Program, crease is proportionally greater than the decrease in total feeding and the National Research Initiative of the USDA Cooperative State Re- rate (i.e., Fig. 6B), then per-nestling feeding rates should increase search, Education and Extension Service, grant number 2005-02817 to with nest predation (Fig. 1D) and could explain faster growth in TEM. Permit numbers for work in Venezuela were DM/0000237 from FONACIT, PA-INP-005-2004 from INPARQUES, and 01-03-03-1147 species with higher predation risk. However, per-nestling feed- from Ministerio del Ambiente. Any use of trade names is for descriptive ing rates decreased with increasing nest predation in all three purposes only and does not imply endorsement by the U.S. Government. sites (Fig. 6C) indicating that provisioning activity is reduced at a faster proportional rate compared with brood size with increasing LITERATURE CITED predation risk. This result follows from theory and other evidence Altwegg, R. 2002. Predator-induced life-history plasticity under time constraints in pool frogs. Ecology 83:2542–2551. suggesting that increased nest predation risk reduces the chances Ardia, D. R. 2006. Geographic variation in the trade-off between nestling of offspring surviving and should favor reduced parental invest- growth rate and body condition in the Tree Swallow. Condor 108:601– ment (effort per offspring) (Ghalambor and Martin 2000, 2001; 611.

EVOLUTION JUNE 2011 1619 THOMAS E. MARTIN ET AL.

Arendt, J. D. 1997. Adaptive intrinsic growth rates: an integration across taxa. Fontaine, J. J., and T. E. Martin. 2006. Parent birds assess nest predation risk Q. Rev. Biol. 72:149–177. and adjust their reproductive strategies. Ecol. Lett. 9:428–434. ———. 2003. Reduced burst speed is a cost of rapid growth in anuran tadpoles: Fontaine, J. J., M. Martel, H. M. Markland, A. Niklison, K. Decker, and problems of autocorrelation and inferences about growth rates. Funct. T. E. Martin. 2007. Testing ecological and behavioral correlates of nest Ecol. 17:328–334. predation. Oikos 116:1887–1894. Arendt, J. D., S. Wilson, and E. Stark. 2001. Scale strength as a cost of rapid Francis, C. M., and D. R. Wells. 2003. The bird community at Pasoh: composi- growth in sunfish. Oikos 93:95–100. tion and population dynamics. Pp. 375–393 in T. Okuda, N. Manokaran, Badyaev, A. V., and T. E. Martin. 2000. Individual variation in growth tra- Y. Matsumoto, K. Niiyama, S. C. Thomas, and P. S. Ashton, eds. Pa- jectories: phenotypic and genetic correlations in ontogeny of the house soh: ecology of a lowland rain forest in southeast Asia. Springer-Verlag, finch (Carpodocus mexicanus). J. Evol. Biol. 13:290–301. Tokyo, Japan. Benard, M. F. 2004. Predator-induced phenotypic plasticity in organisms with Ghalambor, C. K., and T. E. Martin. 2000. Parental investment in two species complex life histories. Annu. Rev. Ecol. Syst. 35:651–673. of nuthatch varies with stage-specific predation risk and reproductive Biancucci, A. L., and T. E. Martin. 2010. Can selection on nest size from nest effort. Anim. Behav. 60:263–267. predation explain the latitudinal gradient in clutch size? J. Anim. Ecol. Ghalambor, C. K., and T. E. Martin. 2001. Fecundity-survival trade-offs and 79:1086–1092. parental risk-taking in birds. Science 292:494–497. Billerbeck, J. M., Lankford, Jr., T. E., and Conover, D. O. 2001. Evo- Grafen, A. 1989. The phylogenetic regression. Philos. Trans. R. Soc. Lond. B lution of intrinsic growth and energy acquisition rates. I. Trade-offs 326:119–157. with swimming performance in Menidia menidia. Evolution 55:1863– ———. 1992. The uniqueness of the phylogenetic regression. J. Theor. Biol. 1872. 156:405–424. Blake, J. G., and B. A. Loiselle. 2008. Estimates of apparent survival rates for Hensler, G. L., and J. D. Nichols. 1981. The Mayfield method of estimating forest birds in eastern Ecuador. Biotropica 40:485–493. nesting success: a model, estimators and simulation results. Wils. Bull. Bosque, C., and M. T. Bosque. 1995. Nest predation as a selective factor in the 93:42–53. evolution of developmental rates in altricial birds. Am. Nat. 145:234– Hinde, C. A., R. A. Johnstone, and R. M. Kilner. 2010. Parent-offspring 260. conflict and coadaptation. Science 327:1373–1376. Brommer, J. E. 2003. Immunocompetence and its costs during development: Johnston, J. P., W. J. Peach, R. D. Gregory, and S. A. White. 1997. Survival an experimental study in blue tit nestlings. Proc. R. Soc. Lond. B rates of tropical and temperate passerines: a Trinidadian perspective. 271:S110–S113. Am. Nat. 150:771–789. Burns, K. J., S. J. Hackett, and N. K. Klein. 2002. Phylogenetic relationships Karr, J. R., J. D. Nichols, M. K. Klimkiewicz, and J. D. Brawn. 1990. Survival and morphological diversity in Darwin’s Finches and their relatives. rates of birds of tropical and temperate forests: will the dogma survive? Evolution 56:1240–1252. Am. Nat. 136:277–291. Case, T. J. 1978. On the evolution and adaptive significance of postnatal Klicka, J., G. Voelker, and G. M. Spellman. 2005. A molecular systematic re- growth rates in the terrestrial vertebrates. Q. Rev. Biol. 53:243–282. vision of the “true thrushes” (Turdinae). Mol. Phylogenet. Evol. 34:486– Chalfoun, A., and T. E. Martin. 2010. Parental investment decisions in re- 500. sponse to ambient nest predation risk versus actual predation on the Kolliker,¨ M., M. W.G. Brinkhof, P.Heeb, P.S. Fitze, and H. Richner. 2000. The prior nest. Condor 112:701–710. quantitative genetic basis of offspring solicitation and parental response Cox, W. A., and T. E. Martin. 2009. Breeding biology of the three-striped in a passerine bird with biparental care. Proc. R. Soc., Lond. B 267:2127– warbler in Venezuela: a contrast between tropical and temperate parulids. 2132. Wils. J. Orn. 121:667–678. Kunz, C., and J. Ekman. 2000. Genetic and environmental components of Crossner, K. A. 1977. Natural selection and clutch size in the European growth in nestling Blue Tits (Parus caeruleus). J. Evol. Biol. 13:199– starling. Ecology 58:885–892. 212. Dial, K. P., E. Greene, and D. J. Irschick. 2008. Allometry of behavior. Trends Lack, D. 1968. Ecological adaptations for breeding in birds. Methuen, London. Ecol. Evol. 23:394–401. Lima, S. 2009. Predators and the breeding bird: behavioral and reproductive Dmitriew, C. M. 2011. The evolution of growth trajectories: what limits growth flexibility under the risk of predation. Biol. Rev. 84:485–513. rate? Biol. Rev. 86:97–116. Linden,´ M., L. Gustafsson, and T. Part.¨ 1992. Selection on fledgling mass in Doligez, B., and J. Clobert. 2003. Clutch size reduction as a response to in- the Collared Flycatcher and the Great Tit. Ecology 73:336–343. creased nest predation rate in the collared flycatcher. Ecology 84:2582– Lloyd, J., and T. E. Martin. 2003. Sibling competition and pre-natal develop- 2588. ment in birds. Proc. R. Soc. Lond. B 270:735–740. Eggers, S., M. Griesser, and J. Ekman. 2005. Predator-induced plasticity in ———. 2004. Nest-site preference and maternal effects on offspring growth. nest visitation rates in the Siberian Jay (Perisoreus infaustus). Behav. Behav. Ecol. 15:816–823. Ecol. 16:309–315. Lyon, B. E., R. D. Montgomerie, and L. D. Hamilton. 1987. Male parental care Eggers, S., M. Griesser, M. Nystrand, and J. Ekman. 2006. Predation risk and monogamy in Snow Buntings. Behav. Ecol. Sociobiol. 20:377–382. induces changes in nest-site selection and clutch size in the Siberian Jay. MacColl, A. D., and B. J. Hatchwell. 2003. Heritability of parental effort in a Proc. R. Soc. Lond. B 273:701–706. passerine bird. Evolution 57:22191–2195. Felsenstein, J. 1985. Phylogenies and the comparative method. Am. Nat. Martin, T. E. 1987. Food as a limit on breeding birds: a life–history perspective. 125:1–15. Annu. Rev. Ecol. Syst. 18:453–487. Ferretti, V., P. E. Llambias, and T. E. Martin. 2005. Life history differences ———. 1992. Interaction of nest predation and food limitation in reproductive between populations of a Neotropical thrush challenge food limitation strategies. Curr. Orn. 9:163–197. theory. Proc. R. Soc. Lond. B 272:769–773. ———. 1995. Avian life history evolution in relation to nest sites, nest pre- Fierro-Calderon,´ K., and T. E. Martin. 2007. Reproductive biology of the dation and food. Ecol. Monogr. 65:101–127. Violet-chested Hummingbird in Venezuela and comparisons with other ———. 1996. Life history evolution in tropical and south temperate birds: tropical and temperate hummingbirds. Condor 109:680–685. what do we really know? J. Av. Biol. 27:263–272.

1620 EVOLUTION JUNE 2011 GROWTH RATE VARIATION

———. 2002. A new view for avian life history evolution tested on an incu- Richner, H., P. Schneiter, and H. Stirnimann. 1989. Life-history consequences bation paradox. Proc. R. Soc. Lond. B 269:309–316. of growth rate depression: an experimental study on carrion crows ———. 2007. Climate correlates of 20 years of trophic changes in a high (Corvus corone corone L.). Funct. Ecol. 3:617–624. elevation riparian system. Ecology 88:367–380. Ricklefs, R. E. 1968. Patterns of growth in birds. Ibis 110:419–451. ———. 2004. Avian life-history evolution has an eminent past: does it have ———. 1976. Growth rates of birds in the humid new world tropics. Ibis a bright future? Auk 121:289–301. 118:179–207. Martin, T. E., and H. Schwabl. 2008. Variation in maternal effects and embry- ———. 1993. Sibling competition, hatching asynchrony, incubation period, onic development rates among passerine species. Philos. Trans. R. Soc. and life span in altricial birds. Curr. Ornithol. 11:199–275. Lond. B 363:1663–1674. Roff, D. A. 1992. The evolution of life histories. Chapman and Hall Publishers, Martin, T. E., and J. V. Briskie. 2009. Predation on dependent offspring: a New York. review of the consequences for mean expression and phenotypic plas- Roff, D. A., V. Remes,ˇ and T. E. Martin. 2005. The evolution of fledging age ticity in avian life history traits. Year Evol. Biol. 2009: Annu. New York in songbirds. J. Evol. Biol. 18:1425–1433. Acad. Sci. 1168:201–217. Sandercock, B. K., S. R. Beissinger, S. H. Stoleson, R. R. Melland, and C. Martin, T. E., P. R. Martin, C. R. Olson, B. J. Heidinger, and J. J. Fontaine. R. Hughes. 2000. Survival rates of a Neotropical parrot: implications 2000a. Parental care and clutch sizes in North and South American birds. for latitudinal comparisons of avian demography. Ecology 81:1351– Science 287:148–1485. 1370. Martin, T. E., J. Scott, and C. Menge. 2000b. Nest predation increases with Sargent, S. 1993. Nesting biology of the Yellow-throated Euphonia: large parental activity: separating nest site and parental activity effects. Proc. clutch size in a Neotropical frugivore. Wils. Bull. 105:285–300. R. Soc. Lond. B 267:2287–2293. Scheuerlein, A., and E. Gwinner. 2006. Reduced nestling growth of East Martin, T. E., R. D. Bassar, S. K. Bassar, J. J. Fontaine, P. Lloyd, H. Math- African Stonechats Saxicola torquata axillaris in the presence of a preda- ewson, A. Niklison, and A. Chalfoun. 2006. Life history and ecological tor. Ibis 148:468–476. correlates of geographic variation in egg and clutch mass among passer- Shkedy, Y., and U. N. Safriel. 1992. Nest predation and nestling growth rate ine species. Evolution 60:390–398. of two lark species in the Negev Desert, Israel. Ibis 134:268–272. Martin, T. E., S. K. Auer, R. D. Bassar, A. M. Niklison, and P. Lloyd. 2007. Sibley, C. G., and J. E. Ahlquist. 1990. Phylogeny and classification of birds: Geographic variation in avian incubation periods and parental influences a study in molecular evolution. Yale Univ. Press, New Haven, CN. on embryonic temperature. Evolution 61:2558–2569. Skutch, A. F. 1949. Do tropical birds rear as many young as they can nourish? Massaro, M., A. Starling-Windhof, J. V. Briskie, and T. E. Martin. 2008. Ibis 91:430–455. Introduced mammalian predators induce adaptive shifts in parental Slagsvold, T. 1982. Clutch size variation in passerine birds: the nest predation behaviour in an endemic New Zealand bird. PLoS One 3:e2331. hypothesis. Oecologia 54:159–169. doi:10.1371/journal.pone.0002331. Soler, J. J., L. deNeve, T. Perez-Contreras,´ M. Soler, and G. Sorci. 2003. Mayfield, H. 1975. Suggestions for calculating nest success. Wils. Bull. Trade-off between immunocompetence and growth in magpies: an ex- 87:456–466. perimental study. Proc. R. Soc. Lond. B 270:241–248. McAdam, A. G., and S. Boutin. 2003. Effects of food abundance on genetic SPSS, Inc. 2009. PASW Statistics 18, Release Version 18.0.3, Chicago, IL. and maternal variation in the growth rate of juvenile red squirrels. J. Starck, J. M., and R. E. Ricklefs, eds. 1998. Avian growth and development. Evol. Biol. 16:1249–1256. Oxford Univ. Press, Oxford, UK. McPeek, M. A. 2004. The growth/predation risk trade-off: so what is the Tello, J. G., R. G. Moyle, D. J. Marchese, and J. Cracraft. 2009. Phylogeny and mechanism? Am. Nat. 163:E88–E111. phylogenetic classification of the tyrant flycatchers, cotingas, manakins, Moreno,J.,J.Vinuela,˜ and J. Belliure. 1998. Effect of brood size on growth in and their allies (Aves: Tyrannides). Cladistics 25:429–467. the chinstrap penguin: a field experiment. J. Field Ornithol. 69:269–275. Thomson, R. L., J. T. Forsman, M. Monkk¨ onen,¨ M. Hukkanen, K. Koivula, S. Naef-Daenzer, B., and L. F. Keller. 1999. The foraging performance of great Rytkonen,¨ and M. Orell. 2006. Predation risk effects on fitness related and blue tits (Parus major and P. caeruleus) in relation to caterpillar measures in a resident bird. Oikos 113:325–333. development, and its consequences for nestling growth and fledging Tremblay, I., D. Thomas, M. M. Lambrechts, J. Blondel, and P. Perret. 2003. weight. J. Anim. Ecol. 68:708–718. Variation in Blue Tit breeding performance. Ecology 84:3033–3043. Nolan, P. M., A. M. Stoehr, G. E. Hill, and K. J. McGraw. 2001. The number of Van Buskirk, J. 2000. The costs of an inducible defense in anuran larvae. provisioning visits by House Finches predicts the mass of food delivered. Ecology 81:2813–2821. Condor 103:851–855. Voelker, G., S. Rohwer, R. C. K. Bowie, and D. C. Outlaw. 2007. Molecular Peach, W. J., D. B. Hanmer, and T. B. Oatley. 2001. Do southern African systematics of a speciose, cosmopolitan songbird genus: defining the songbirds live longer than their European counterparts? Oikos 93:235– limits of, and relationships among, the Turdus thrushes. Mol. Phylogenet. 249. Evol. 42:422–434. Pettifor, R. A., C. M. Perrins, and R. H. McCleery. 2001. The individual opti- Walsh, M. R., and D. N. Reznick. 2008. Interactions between the direct and mization of fitness: variation in reproductive output, including clutch indirect effects of predators determine life history evolution in a killifish. size, mean nestling mass and offspring recruitment, in manipulated Proc. Natl. Acad. Sci. USA 105:594–599. broods of great tits Parus major. J. Anim. Ecol. 70:62–79. Williams, G. C. 1966. Adaptation and natural selection. Princeton Univ. Press, Purvis, A., and A. Rambaut. 1995. Comparative analysis by independent NJ. contrasts (CAIC): an Apple Macintosh application for analysing com- Yuri, T., and D. P. Mindell. 2002. Molecular phylogenetic analysis of Fringill- parative data. Comp. Appl. Biol. Sci. 11:247–251. idae, ‘‘New World nine-primaried oscines’’ (Aves: Passeriformes). Mol. Remes,ˇ V., and T. E. Martin. 2002. Environmental influences on the evolution Phylogenet. Evol. 23:229–243. of growth and developmental rates in Passerines. Evolution 56:2505– 2518. Associate Editor: M. Webster

EVOLUTION JUNE 2011 1621 THOMAS E. MARTIN ET AL.

Supporting Information The following supporting information is available for this article:

Appendix S1. Identities of species and their taxonomic family in each of the three sites. Appendix S2. Structural equation modeling methods and results.

Supporting Information may be found in the online version of this article.

Please note: Wiley-Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.

1622 EVOLUTION JUNE 2011