Frequency and density-dependent selection on life-history strategies - a field experiment

F equency and Densi y-Dependen Selec ion on Li e- His o y S a egies – A Field Expe imen Tapio Mappes 1 *, Minna Koi ula 2 , Esa Koskela 1 , Tuula A. Oksanen 1 , Tiina Sa olainen 1 , Ba y Sine o 3 1Depa men o Biological and En i onmen al Science, Uni e si y o Jy a ¨skyla ¨, Jy a ¨skyla ¨, Finland, 2MTT Bio echnology and Food Resea ch, Jokioinen, Finland, 3Depa men o Ecology and E olu iona y Biology, Uni e si y o Cali o nia San a C uz, San a C uz, Cali o nia, Uni ed S a es o Ame ica Abs ac Nega i e equency-dependence, which a o s a e geno ypes, p omo es he main enance o gene ic a iabili y and is o in e es as a po en ial explana ion o gene ic di e en ia ion. Densi y-dependen selec ion may also p omo e cyclic changes in equencies o geno ypes. He e we show e idence o bo h densi y-dependen and nega i e equency-dependen selec ion on opposi e li e-his o y ac ics (low o high ep oduc i e e o , RE) in he bank ole (Myodes gla eolus). Densi y- dependen selec ion was e iden among he emales wi h low RE, which we e especially a o ed in low densi ies. Ins ead, bo h nega i e equency-dependen and densi y-dependen selec ion we e shown in emales wi h high RE, which we e mos success ul when hey we e a e in high densi ies. Fu he mo e, selec ion a he indi idual le el a ec ed he equencies o ac ics a he popula ion le el, so ha he equency o he a e high RE ac ic inc eased signi ican ly a high densi ies. We hypo hesize ha hese wo selec ion mechanisms (densi y- and nega i e equency-dependen selec ion) may p omo e gene ic a iabili y in cyclic mammal popula ions. Ne e heless, i emains o be de e mined whe he he o igin o gene ic a iance in li e-his o y ai s is causally ela ed o densi y a ia ion (e.g. popula ion cycles). Ci a ion: Mappes T, Koi ula M, Koskela E, Oksanen TA, Sa olainen T, e al (2008) F equency and Densi y-Dependen Selec ion on Li e-His o y S a egies – A Field Expe imen . PLoS ONE 3(2): e1687. doi:10.1371/jou nal.pone.0001687 Edi o : Tom T egenza, Uni e si y o Exe e , Uni ed Kingdom Recei ed Oc obe 11, 2007; Accep ed Janua y 17, 2008; Published Feb ua y 27, 2008 Copy igh : ß2008 Mappes e al. This is an open-access a icle dis ibu ed unde he e ms o he C ea i e Commons A ibu ion License, which pe mi s un es ic ed use, dis ibu ion, and ep oduc ion in any medium, p o ided he o iginal au ho and sou ce a e c edi ed. Funding: Financial suppo ed by he Academy o Finland (g an no. 206091, 118603, 109165 o T.M; 100143, 78777, 103148, 115961 o E.K.) and Cen e o Excellence o E olu iona y Resea ch o he Academy o Finland. Compe ing In e es s: The au ho s ha e decla ed ha no compe ing in e es s exis . *E-mail: [email p o ec ed] In oduc ion A undamen al p oblem in e olu iona y biology is o ind mechanisms main aining addi i e gene ic a ia ion in na u al popula ions [1,2]. Acco ding o e olu iona y heo y, selec ion should educe gene ic a ia ion especially in he ai s ha a e closely associa ed wi h i ness [1]. Howe e , many species s ill exhibi la ge gene ic a ia ion in i ness- ela ed li e-his o y ai s [3–6]. A leas i e selec ion mechanisms a e hypo hesized o main ain gene ic a ia ion in na u e: mu a ion-selec ion balance, he e osis, an agonis ic pleio opy, nega i e equency-dependen selec ion (ad an age o a e geno ype) and en i onmen al he e ogenei y [2,7]. Many heo e ical analyses ha e ocused on he la e wo selec ion mechanisms (e.g. ecen ly [8]), and hese models p edic s ha , o example, nega i e equency-dependen selec ion can be mo e common in na u al popula ions as p e iously ecognized. S ill, all empi ical s udies exclusi ely consis o polymo phic popula ions, whe e gene ic colou mo phs a e e iden (e.g in plan s, ish and ep iles) [9–11]. In polymo phic sys ems i has been mo e ob ious o es whe he he i ness o indi iduals depends on hei neighbou s’ geno ype. Among he o ganisms wi hou isible polymo phism, bu which s ill ha e la ge gene ic a ia ion in impo an i ness ai s (e.g. in li e-his o y ai s), nega i e equency-dependen selec ion (ad an age o a e geno ype) has no ye ecei ed wide a en ion. He e we aimed o es empi ically wo selec ion mechanisms: nega i e equency-dependen selec ion and en i onmen al he - e ogenei y (densi y-dependen selec ion), which a e p edic ed o main ain gene ic a ia ion in na u al popula ions [12–18]. Selec ion was s udied in he bank ole (Myodes gla eolus), a small mammal wi h high pheno ypic and gene ic a ia ion in li e-his o y ai s, such as ep oduc i e e o , o sp ing size and numbe [19]. Signi ican empo al a ia ion o densi y-dependen selec ion is p omo ed by la ge seasonal [20] and 3–4 yea cyclic densi y a ia ion in his species [21]. Fu he mo e, compe i ion be ween e i o ial bank ole emales is a majo mechanism de e mining hei b eeding success which, especially a high densi ies, leads o la ge a ia ion in ela i e i ness o indi iduals [22–24]. Toge he hese selec i e en i onmen s could acili a e he o igin and exis ence o opposi e li e-his o y ac ics whose success would depend bo h on he cu en en i onmen and he equency o opposi e ac ic in he popula ion. This idea is suppo ed by s udies wi h side-blo ched liza ds (U a s ansbu iana) and common liza ds (Lace a i ipa a), whe e nega i e equency-dependen selec ion and densi y a ia ion ha e been shown o con ibu e o gene ic cycles d i en by al e na i e li e-his o y s a egies [25–27]. He e we ocused on one cen al li e-his o y ai , ep oduc i e e o (RE) o emales, and expe imen ally s udied whe he opposi e ac ics (high s. low RE) would be a o ed by di e en selec ion p essu es. Bo h he equencies ( a e s. common) and densi ies (low s. high) o he RE ac ics we e manipula ed in la ge enclosed popula ions, whe e hei ela i e su i al and b eeding success we e moni o ed o e he b eeding season. Ou aim was o expe imen ally es whe he condi ions occu ing in cyclic small mammal popula ions could acili a e he o igin o opposi e li e- his o y ac ics which would hen p omo e la ge gene ic a ia ion obse ed in se e al ai s. PLoS ONE | www.plosone.o g 1 Feb ua y 2008 | Volume 3 | Issue 2 | e1687 Resul s When s udying ju enile ec ui men o he adul popula ion, we ound ha bo h he numbe o o sp ing weaned and p opo ion o su i ing o sp ing we e ela ed o he le el o ep oduc i e e o and he equency and densi y o al e na i e RE ac ics in adul emales (Th ee-way in e ac ions: F 1,64 =7.16, P= 0.009; F 1,64 =7.82, P= 0.007, espec i ely). This indica es ha he success o RE ac ics di e ed acco ding o he equencies and densi ies, and so in u he analyses RE ac ics we e analyzed sepa a ely (Table 1 and 2). In pa icula , emales wi h low RE we e a o ed a low densi y bu RE equency had no signi ican e ec on hei b eeding success (Fig. 1A, S1, Table 1, ). Females wi h high RE we e mos success ul when hey we e a e in he popula ion and a high densi y (Fig. 1B, S1, Table 1). Analyses o selec ion g adien s suppo hese esul s as hey indica e di ec ional densi y-dependen selec ion owa ds highe ep oduc i e e o only among a e ac ics (Fig. 2). So, nega i e equency-dependen selec ion on highe ep oduc i e e o wo ks e ec i ely in high densi y popula ions. Toge he , hese indings sugges ha low RE emales a e success ul only in low popula ion densi ies. Acco ding o li e-his o y heo y, he success o al e na i e RE ac ics can also be shaped by he ade-o be ween RE and ep oduc i e cos s. Indeed, signi ican equency-dependen su i al cos s we e associa ed wi h high RE ac ics, especially when hey we e common in he popula ion (Table 2). O no e, he main esul s o he densi y- and equency-dependen e ec s on b eeding success (Fig. 1A,B, S1, Table 1) we e no biased by su i al cos s, as he su i al o emales was no ye a ec ed by RE ac ic o manipula ions o equencies and densi ies du ing hei i s b eeding (Linea logi model, G,0.090, P.0.663 o all main e ec s and in e ac ions). Such su i al cos s we e mani es ed in la e b eeding episodes. The su i al o mo he s and hei o sp ing we e moni o ed un il he end o he b eeding season o es ima e he changes in equencies o RE ac ics a he popula ion le el. F equency o high RE ac ics inc eased signi ican ly when densi ies we e high and ini ial equency was low ( a e ac ic) ( = 23.7, d =2, P= 0.002), and simul aneously he equency o low RE emales (common and high densi y) dec eased signi ican ly (Fig. 3). Discussion Ou esul s indica e clea nega i e equency-dependen and densi y-dependen selec ion on di e en b eeding ac ics in bank oles. Females wi h low ep oduc i e e o we e especially a o ed in low densi ies. Mo eo e , emales wi h high ep oduc i e e o we e mos success ul when hey we e a e in high densi y popula ions. The di e en successes o RE ac ics we e con i med by bo h he b eeding success and su i al o mo he s. When s udying how selec ion a he indi idual le el a ec ed he equencies o ac ics a he popula ion le el, we ound ha he equency o he a e high e o ac ic inc eased signi ican ly in high densi ies (Fig. 3). Acco ding o hese esul s, we hypo hesize ha a e he c ash o a ole popula ion, low RE emales ha e he highes b eeding success un il he popula ion size inc eases, a which ime high RE emales ob ain a a e ac ic ad an age. Ano he c ucial phase o he li e his o y occu s du ing and a e he win e c ash phase, which migh be bene icial o indi iduals wi h low RE. This idea p edic s ha he equencies o opposi e b eeding ac ics change acco ding o he seasonal o mul i-annual densi y a ia ions in ole popula ions. Nega i e equency-dependen selec ion e iden ly shows com- pe i ion be ween ac ics [28]; he e i is in aspeci ic compe i ion be ween he b eeding ac ics. In bank oles, high RE emales Table 1. The e ec s o equency and densi y on numbe o o sp ing weaned, and p opo ion o o sp ing su i ing un il weaning in di e en ac ics o ep oduc i e e o . Low ep oduc i e e o Numbe o o sp ing weaned P opo ion o o sp ing su i ing d 1 d 2 FPd 1 d 2 FP Indi idual le el F equency 1 18.2 0.19 0.669 1 34.0 0.01 0.935 Densi y 1 18.2 7.24 0.015 1 34.0 11.04 0.002 F eq * Den 1 18.2 2.58 0.126 1 34.0 2.13 0.154 Popula ion le el F equency 1 9 0.23 0.642 1 9 0.01 0.919 Densi y 1 9 8.89 0.015 1 9 17.72 0.002 F eq * Den 1 9 3.17 0.109 1 9 3.42 0.098 High ep oduc i e e o Numbe o o sp ing weaned P opo ion o o sp ing su i ing d 1 d 2 FPd 1 d 2 FP Indi idual le el F equency 1 30 2.19 0.149 1 30 4.35 0.046 Densi y 1 30 0.08 0.783 1 30 0.01 0.930 F eq * Den 1 30 4.53 0.042 1 30 6.79 0.014 Popula ion le el F equency 1 9 2.33 0.161 1 9 3.44 0.097 Densi y 1 9 0.82 0.781 1 9 0.01 0.939 F eq * Den 1 9 4.82 0.056 1 9 5.38 0.046 No es: The analyses o gene alized linea mixed models a e pe o med bo h a he indi idual le el (indi idual alues o med he dependen a iables) and a he popula ion le el (popula ion means o med he dependen a iables). A he indi idual le el, he andom e ec o popula ion (enclosu e) is included in he models (Es ima e,0.006, P.0.989 in all cases). d 1 = nume a o d , d 2 = denomina o d doi:10.1371/jou nal.pone.0001687. 001 F equency-Dependen Selec ion PLoS ONE | www.plosone.o g 2 Feb ua y 2008 | Volume 3 | Issue 2 | e1687 seems o be dominan o e he low RE emales. As he compe i ion o space is he mos impo an ac o a ec ing b eeding success o e i o ial emale bank oles [22,29–31], he high RE emales migh ha e a g ea e abili y o occupy and de end hei e i o ies especially agains low RE emales in high densi ies. In small mammals, e i o iali y unc ions o de end ood esou ces and/o o sp ing agains in an icide [32,33]. The ask o u u e s udies is o de e mine he impo ance o hese wo unc ions o he RE ac ics in di e en densi ies. In he cu en s udy, emales wi h high ep oduc i e e o ha e lowe su i al as p edic ed by he heo y o ep oduc i e cos s [34,35]. These esul s a e consis en wi h hose o ou ea lie expe imen al s udies, which epo a ade-o be ween ep oduc- i e e o (li e size) and mo he su i al [20,36]. In e es ingly, he ep oduc i e cos s o a mo he can be a ec ed by o he mo he s as well as hei b eeding ac ic in he popula ions. High su i al cos s o high RE emales a e e iden especially when hey compe e agains he mo he s wi h he same (common) ac ic (Table 2). This could be caused by la ge ood equi emen s o high RE emales. Unde in ense compe i ion hey migh no be able o alloca e bo h o ep oduc ion and hei own su i al. The densi y luc ua ion causing i ness di e ences in RE ac ics can be annual o mul i-annual. I has been hypo hesized ha e en he popula ion dynamics (e.g. popula ion cycles) in small mammals a e de e mined by he luc ua ions in geno ypes [37]. This hypo hesis o in insic egula ion has no ecei ed clea e idence, and so he dynamics o cyclici y (e.g. popula ion c ashes) a e mo e e iden ly a ec ed by ex e nal ac o s e.g. p eda o s and/o ood [38–40] S ill, he ‘‘so ’’ hypo hesis o Chi y, ha gene ic a ia ion is a consequence and no a cause o popula ion cycles, could be alid. Howe e , e en a e ex ensi e esea ch, his idea is suppo ed only by a ew gene ic s udies in cyclic mammals [41,42] In conclusion, we would like o emphasize ha i emains o be de e mined whe he he o igin o gene ic a iance in li e-his o y ai s is causally ela ed o densi y a ia ion (e.g. popula ion cycles). Ne e heless, ou s udy demons a es ha selec ion on al e na i e Figu e 1. Numbe o o sp ing weaned o low (a) and high RE emales (b) in di e en densi ies and equencies. Closed ci cles: a e ac ic; Open ci cles: common ac ic in he popula ions. doi:10.1371/jou nal.pone.0001687.g001 Table 2. Su i al di e ences be ween he high and low RE emales and whe he he ac ic was a e o common in he popula ion. Pe cen age (N) o emales su i ing o he end o b eeding season Tac ic Ra e Common All Low RE 42.9 (7) 63.0 (27) 58.8 (34) High RE 66.7 (9) 19.0 (21) 33.3 (30) Indi idual le el d 1 d 2 FP RE ac ic 1 57 0.89 0.348 F equency 1 57 0.75 0.391 Densi y 1 26.7 0.57 0.458 RE ac ic * F equency 1 23.6 6.02 0.022 RE ac ic *Densi y 1 57 0.29 0.595 F equency * Densi y 1 57 1.64 0.206 Popula ion le el d 1 d 2 FP RE ac ic 1 15 0.40 0.536 F equency 1 15 0.40 0.536 Densi y 1 15 0.24 0.631 RE ac ic * F equency 1 15 4.79 0.045 RE ac ic *Densi y 1 15 0.09 0.772 F equency * Densi y 1 15 0.94 0.347 No es: The analyses o gene alized linea mixed models a e pe o med bo h a he indi idual le el (indi idual alues o med he dependen a iable) and popula ion le el (popula ion means o med he dependen a iable). A he indi idual le el, he andom e ec o popula ion (enclosu e) is included in he models (Es ima e = 0.046, P= 0.339). d 1 = nume a o d , d 2 = denomina o d doi:10.1371/jou nal.pone.0001687. 002 F equency-Dependen Selec ion PLoS ONE | www.plosone.o g 3 Feb ua y 2008 | Volume 3 | Issue 2 | e1687 li e-his o y s a egies would a ise om he ac ion o densi y and equency cycles. Ma e ials and Me hods S udy species The s udy species, he bank ole (Myodes gla eolus), is a small oden species common in no he n Eu ope [43]. The main habi a s a e o es s and ields, and he die consis s o o bs, shoo s, seeds, be ies and ungi [44]. The densi y o b eeding emales is limi ed h ough hei e i o iali y [29], and in high densi ies he ma u a ion o young emales is supp essed by social in e ac ion among emales [45]. The pa e ns and ampli ude o densi y a ia ion show conside able geog aphical a iabili y, and bo h s able and cyclic popula ions a e ound [46]. In ou s udy a ea, emales gi e bi h o a maximum o ou li e s du ing he b eeding season, which las s om la e Ap il o Sep embe . In addi ion o la ge pheno ypic [20] and gene ic a ia ion in li e size (2–10) and o sp ing size (1.3–2.5 g), a nega i e gene ic co ela ion also exis s be ween hese ai s [19]. P egnancy las s o 19–20 days and pups a e weaned un il he age o h ee weeks [47]. Bank oles ha e good appabili y, and hey a e no sensi i e o dis u bance, which allows moni o ing popula ions by li e- apping. S udy p ocedu es Expe imen al animals o igina ed om a i icial selec ion lines a he Uni e si y o Jy a¨skyla¨. Founde animals o he selec ion lines we e caugh in cen al Finland (62u379N, 26u209E). In he unca ion selec ion expe imen , wo lines we e selec ed acco ding o li e size o emales (see de ails Sch ode us e al. 2007, submi ed manusc ip ). The selec ion expe imen was based on he be ween amily selec ion p ocedu e [48], which p e en s he possibili y o inb eeding. Rep oduc i e e o o emales is de e mined he e by he o mula o [49]: RE = (li e size * mean o sp ing mass 0.75 )/ mo he s pos -pa um mass 0.75 ). In his o mula ene gy equi e- men s o p oduce o sp ing a e calcula ed ela i e o he allome ic equi emen o he mo he (assuming s anda d me abolism inc eases o he 0.75 powe o mass o mammals [50]). The s udy s a ed by ma ing a ac ion o he emales (n= 91) ob ained om he selec ion expe imen wi h andomly chosen males o he same line o emale (low RE: mean6SE = 0.6760.02, n= 49; High RE: mean6SE = 0.8460.02, n= 42). We u he inc eased he di e ence be ween he wo g oups by choosing only he emales wi h lowes RE (n= 39) and highes RE (n= 33) o he p esen expe imen . All emales we e a he same age and in simila ep oduc i e s a e. Cha ac e is ic o hese mo he s and hei o sp ing and s a is ical es s a e p esen ed in he Table 3. These 72 emales ga e bi h in he labo a o y in ea ly July, and a e ha he emales wi h hei new-bo n indi idually ma ked pups we e eleased o 13 la ge ou doo enclosu es (each 0.2 ha) [47]. None o he mo he s abandoned hei pups du ing he eleasing p ocess. Fou males we e in oduced o each enclosu e o keep emales in ep oduc i e condi ion. These males migh ma e wi h he emales, bu we we e no able o measu e he success o he possible subsequen b eeding e en s. In he enclosu es, bo h he densi ies (4 o 8 emales/0.2 ha) and equencies (1:3 (4 enclosu es); 3:1 (4), 2:6 (2), 6:2 (3)) o di e en ep oduc i e ac ics (low o high RE) we e manipula ed (see he design in Table 4). The choice o 4 and 8 emales pe enclosu e o low and high expe imen al densi ies espec i ely, was based on ou ea lie s udies showing ha 4–6 emales can gain a e i o y and b eed simul aneously in he same enclosu e [51]. The b eeding success o all 72 emales was s udied du ing one b eeding in July. B eeding success was de e mined by moni o ing he numbe and he p opo ion o o sp ing su i ing o he weaning age (abou 25 days old). Fo moni o ing indi idual oles, 20 mul i-cap u e li e aps we e dis ibu ed in each enclosu e in a 465 a ay wi h 10 m be ween he ap s a ions. Mo he s and Figu e 2. Densi y-dependen selec ion o ep oduc i e e o among a e ac ics. In he analyses o selec ion g adien s, he ela i e i ness (numbe o o sp ing weaned/mean numbe o o sp ing weaned in he popula ion) was es ima ed in ela ion o s anda dized ep oduc i e e o ((RE i –RE mean )/RE SD ). Selec ion g adien s indica e di ec ional selec ion owa ds highe ep oduc i e e o in high densi ies (closed ci cles and solid line: b6SE = 0.6060.15, P= 0.004) bu no in low densi ies (open ci cles and dashed line: b6SE = 20.0860.21, P= 0.726) (S anda dized RE * Densi y in e ac ion: F 1,14 = 7.34, P= 0.017). The selec ion g adien among common ac ics was non-signi ican (b6SE = 0.3360.44, P= 0.453) and densi y-indepen- den (S anda dized RE * Densi y in e ac ion: F 1,50 = 0.02, P= 0.904). doi:10.1371/jou nal.pone.0001687.g002 Figu e 3. Change in equencies o low and high RE ac ics in he di e en ea men popula ions du ing he b eeding season. F equency o high RE ac ics inc eased signi ican ly when densi ies we e high and ini ial equency was low ( a e ac ic), which simul aneously dec eased he equency o low RE (common ac ic). Open ci cles: low RE; Closed ci cle: high RE. N = numbe o eplica es (popula ions). doi:10.1371/jou nal.pone.0001687.g003 F equency-Dependen Selec ion PLoS ONE | www.plosone.o g 4 Feb ua y 2008 | Volume 3 | Issue 2 | e1687 o sp ing we e apped a weaning and a he end o he expe imen (o sp ing abou h ee mon hs old) (see de ails o apping p ocedu e [24,52]). Su i al o only 64 emales was de e mined om he beginning o he expe imen (ea ly July) o he end o he expe imen (ea ly Oc obe ), as all he indi iduals in wo enclosu es escaped be o e he end o he expe imen . The emales escaped a e he i s li e was weaned, so we we e able o include all emales o he analyses o he b eeding success (Table 1 and Fig. 1A,B). Only he success ul enclosu es (see numbe o eplica es in Fig. 3) we e included in o he analyses o equency changes in he popula ion le el (see below). No o he indica ions o unsuccess ul eplica es we e ound du ing he expe imen . Selec ion o ep oduc i e e o was s udied using he analyses o selec ion g adien s [48,53]. Selec ion g adien s (b) we e es ima ed om he linea eg ession coe icien s o ela i e i ness on he s anda dized ai . He e he ela i e i ness (numbe o o sp ing weaned/mean numbe o o sp ing weaned in he popula ion) was es ima ed in ela ion o s anda dized ep oduc i e e o ((RE i –RE mean )/RE SD ). The su i al o mo he s (Table 2) and hei o sp ing was moni o ed un il he end o he b eeding season o es ima e he changes in equencies o RE ac ics a he popula ion le el (Fig. 3). Change in equency was de e mined by he equency o he ce ain b eeding ac ic (su i ed mo he s and hei o sp ing) a he end o b eeding season minus he ini ial equency o he b eeding ac ic in he enclosu e. S a is ical analyses The da a we e analysed using SPSS 14.0 o Windows and SAS e sion 9.1 so wa e. In he gene alized linea mixed model analysis (GLMM), he numbe o o sp ing weaned, p opo ion o o sp ing su i ing o mo he su i al we e explained by RE ac ic, equency, densi y and hei in e ac ions. The analyses o GLMM we e pe o med bo h a he indi idual le el (indi idual alues o med he dependen a iables) and popula ion le el (popula ion means o med he dependen a iables). A he indi idual le el, GLMM allows o he popula ion (enclosu e) o be included as a andom e ec in he models [54]. Suppo ing In o ma ion Figu e S1 Found a : doi:10.1371/jou nal.pone.0001687.s001 (5.22 MB TIF) Acknowledgmen s We hank Johanna Mappes, Suzanne C. Mills, Mikael Mo¨kko¨nen, Tanja Poikonen and F. S ephen Dobson o aluable commen s. Au ho Con ibu ions Concei ed and designed he expe imen s: BS TM MK EK TO. Pe o med he expe imen s: TM MK EK TO TS. Analyzed he da a: TM. Con ibu ed eagen s/ma e ials/analysis ools: TM. W o e he pape : TM. Re e ences 1. P ice T, Schlu e D (1991) On he low he i abili y o li e-his o y ai s. E olu ion 45: 853–861. 2. Ro DR (2002) Li e his o y e olu ion. U.S.A.: Sinaue . 3. Gus a sson L (1986) Li e ime ep oduc i e success and he i abili y: empi ical suppo o Fishe ’s undamen al heo em. Am Na 128: 761–764. 4. Mousseau TA, Ro DA (1987) Na u al selec ion and he he i abili y o i ness componen s. He edi y 59: 181–198. 5. Me ila¨ J, Sheldon BC (2000) Li e ime ep oduc i e success and he i abili y in na u e. Ame ican Na u alis 155: 301–310. 6. McClee y RH, Pe i o RA, A mb us e P, Meye K, Sheldon BC, e al. (2004) Componen s o a iance unde lying i ness in a na u al popula ion o he g ea i Pa us majo . Ame ican Na u alis 164: E62–E72. 7. Ro DA (1992) The e olu ion o li e his o ies. New Yo k, London: Chapman & Hall. Table 3. Cha ac e is ics (mean6SE) o wo emale ac ics (low o high ep oduc i e e o , RE). Sou ce o a iance ( F 1,64 ) Low RE ( n = 39) High RE ( n = 33) RE ac ic F equency Densi y Rep oduc i e e o 0.6160.02 0.8860.02 59.56 *** 0.11 ns 1.91 ns Li e size 3.8260.14 6.1560.18 76.36 *** 0.10 ns 3.52 ns Mean body mass o o sp ing (g) 1.9760.03 1.8460.02 10.60 ** 0.02 ns 2.07 ns Mean head wid h o o sp ing (mm) 8.2860.05 8.0960.04 6.81 * 0.00 ns 0.07 ns Pos -pa um body mass o mo he (g) 22.460.4 24.660.4 6.87 * 0.01 ns 0.27 ns Pos -pa um head wid h o mo he (g) 13.860.07 13.860.06 0.07 ns 1.34 ns 0.10 ns Random assignmen o he di e en manipula ion g oups ( equency and densi y) is es ed by h ee-way ANOVA. All possible wo and h ee-way in e ac ions we e non- signi ican (P.0.11). ***P,= 0.001, **P,0.01, *P,0.05 doi:10.1371/jou nal.pone.0001687. 003 Table 4. The design o he expe imen . Low densi y Low densi y o al High densi y High densi y o al O e all o al Ra e Low RE +Common High RE 1+3 (4) 4+12 2+6(2) 4+12 Ra e High RE +Common Low RE 1+3 (4) 4+12 2+6 (3) 6+18 To al (Low Re+High RE) 16+16 = 32 22+18 = 40 32+40 = 72 Numbe o emales pe enclosu e, numbe o enclosu es (in pa en hesis) and o al numbe o indi iduals in each ea men . doi:10.1371/jou nal.pone.0001687. 004 F equency-Dependen Selec ion PLoS ONE | www.plosone.o g 5 Feb ua y 2008 | Volume 3 | Issue 2 | e1687 8. T o e MV, Spence HG (2007) F equency-dependen selec ion and he main enance o gene ic a ia ion: Explo ing he pa mne e space o he mul iallelic pai wise in e ac ion model. Gene ics 176: 1729–1740. 9. Sine o B, Li ely CM (1996) The ock-pape -scisso s game and he e olu ion o al e na i e male ep oduc i e s a egies. Na u e 380: 240–243. 10. An ono ics J, Ells and NC (1984) Expe imen al S udies o he E olu iona y Signi icance o Sexual Rep oduc ion .1. A Tes o he F equency-Dependen Selec ion Hypo hesis. E olu ion 38: 103–115. 11. Olendo R, Rodd FH, Punzalan D, Houde AE, Hu C, e al. (2006) F equency-dependen su i al in na u al guppy popula ions. 441: 633–636. 12. Bulme MG (1985) The ma hema ical heo y o quan i a i e gene ics. Ox o d: Cla edon p ess. 13. Mani GS, Cla ke BS, Shel om PR (1990) A model o quan i a i e ai s unde equency-dependen balancing selec ion. P oceedings o Royal Socie y o London, Biological Sciences 240: 15–28. 14. Sine o B (2001) Runaway social games, gene ic cycles d i en by al e na i e male and emale s a egies, and he o igin o mo phs. Gene ica 112–113: 417–434. 15. Rue le C, Van Doo en TJM, Me z JAJ (2003) Adap i e walks on changing landscapes: Le ins’ app oach ex ended. Theo e ical Popula ion Biology 65: 165–178. 16. Kopp M, He misson J (2006) The e olu ion o gene ic a chi ec u e unde equency-dependen dis up i e selec ion. E olu ion 60: 1537–1550. 17. Ho¨p ne R, B odda K (2006) A s ochas ic model and a unc ional cen al limi heo em o in o ma ion p ocessing in la ge sys ems o neu ons. Jou nal o Ma hema icl Biology 52: 439–457. 18. Bu ge R, Gimel a b A (2004) The e ec s o in aspeci ic compe i ion and s abilizing selec ion on a polygenic ai . Gene ics 167: 1425–1443. 19. Mappes T, Koskela E (2004) Gene ic basis o he ade-o be ween o sp ing numbe and quali y in he bank ole. E olu ion 58: 645–650. 20. Koi ula M, Koskela E, Mappes T, Oksanen TA (2003) Cos s o ep oduc ion in he wild: manipula ion o ep oduc i e e o in he bank ole. Ecology 84: 398–405. 21. Ko pima¨ki E, No dahl K, Hui u O, Klemola T (2005) P eda o -induced synch ony in popula ion oscilla ions o coexis ing small mammal species. P oceedings o he Royal Socie y B: Biological Sciences 272: 193–202. 22. Jonsson P, Ha ikainen T, Koskela E, Mappes T (2002) De e minan s o ep oduc i e success in oles: space use in ela ion o ood and li e size manipula ion. E olu iona y Ecology 16: 455–467. 23. Koskela E, Mappes T, Ylonen H (1999) Expe imen al manipula ion o b eeding densi y and li e size: e ec s on ep oduc i e success in he bank ole. Jou nal o Animal Ecology 68: 513–521. 24. Oksanen TA, Koi ula M, Koskela E, Mappes T (2007) The cos o ep oduc ion induced by body size a bi h and b eeding densi y. E olu ion 61: 2822–2831. 25. Sine o B, S ensson E, Comendan T (2000) Densi y cycles and an o sp ing quan i y and quali y game d i en by na u al selec ion. Na u e 406: 985–988. 26. Sine o B, Heulin B, Su ge -G oba Y, Clobe J, Miles D, e al. (2007) Models o Densi y Dependen Genic Selec ion and a New Rock-Pape -Scisso s Social Sys em. The Ame ican Na u alis 170: 663–680. 27. Sine o B, Bleay C, Adamopoulou C (2001) Social causes o co ela ional selec ion and he esolu ion o a he i able h oa colo polymo phism in a liza d. E olu ion 55: 2040–2052. 28. Mayna d Smi h J (1982) E olu ion and he Theo y o Games. Camb idge: Camb idge Uni e si y P ess. 224 p. 29. Bujalska G (1970) Rep oduc ion s abilizing elemen s in island popula ions o Cle h ionomys gla eolus. Ac a The iologica 15: 381–412. 30. Koskela E (1997) Te i o ial beha iou and ep oduc i e success o bank ole Cle h ionomys gla eolus emales. Jou nal o Animal Ecology 66: 341–349. 31. Koskela E, Mappes T, Ylo¨nen H (1999) Expe imen al manipula ion o b eeding densi y and li e size: e ec s on ep oduc i e success in he bank ole. Jou nal o Animal Ecology 68: 513–521. 32. Os eld RS (1990) The ecology o e i o iali y in small mammals. T ends in Ecology and E olu ion 5: 411–415. 33. Wol JO (1993) Why a e emales small mammals e i o ial? Oikos 68: 364–370. 34. Reznick D (1985) Cos o ep oduc ion: an e alua ion o empi ical e idence. Oikos 44: 257–267. 35. Williams GC (1966) Na u al selec ion, he cos o ep oduc ion, and a e inemen o Lack’s p inciple. Ame ican Na u alis 100: 687–690. 36. Oksanen TA, Koskela E, Mappes T (2002) Ho monal manipula ion o o sp ing numbe : ma e nal e o and ep oduc i e cos s. E olu ion 56: 1530–1537. 37. Chi y D (1967) The na u al selec ion o sel - egula o y beha iou in animal popula ions. P oc Ecol Soc Aus 2: 51–78. 38. Hanski I, Hen onen H, Ko pimaki E, L. O, Tu chin P (2001) Small- oden dynamics and p eda ion. Ecology 82: 1505–1520. 39. Tu chin P, Ba zli GO (2001) A ailabili y o ood and he popula ion dynamics o a icoline oden s. Ecology 82: 1521–1534. 40. Hui u O, Jokinen I, Ko pima¨ki E, Koskela E, Mappes T (2006) Phase dependence in win e physiological condi ion o cyclic oles. Oikos (in p ess). 41. Wo´jcik AM, Polly D, Siko ski MD, Wo´jcik J (2006) Selec ion in cycling popula ion: di en ial esponse among skele al ai s. E olu ion 60: 1925–1935. 42. Lidicke WZJ, Wol JO, Mow ey RA (2000) Gene ic change in a cyclic popula ion o snowshoe ha es. Ecoscience 7: 247–255. 43. S ense h NC (1985) Geog aphic dis ibu ion o Cle h ionomys species. Annales Zoologici Fennici 22: 215–219. 44. Hansson L (1979) Condi ion and die in ela ion o habi a in bank oles Cle ionomys gla eolus: popula ion and communi y app oach? Oikos 33: 55–63. 45. K uczek M, Ma chlewska-Koj A (1986) Pube y delay o bank ole emales in high densi y popula ion. Biol Rep od 35: 537–541. 46. Hansson L, Hen onen H (1985) G adien s in densi y a ia ions o small oden s: he impo ance o la i ude and snow co e . Oecologia 67: 394–402. 47. Mappes T, Koskela E, Ylo¨nen H (1995) Rep oduc i e cos s and li e size in he bank ole. P oceedings o Royal Socie y o London, Biological Sciences 261: 19–24. 48. Falcone DS, Mackay TFC (1996) In oduc ion o quan i a i e gene ics. Ha low, UK.: Longman Sci. and Tech. 49. Milla JS (1977) Adap i e ea u es o mammalian ep oduc ion. E olu ion 31: 370–386. 50. Kleibe M (1961) The i e o li e. New Yo k: John Wiley & Sons Publ. 51. Oksanen TA, Jokinen I, Koskela E, Mappes T, Vilpas H (2003) Manipula ion o o sp ing numbe and size, bene i s o la ge body size a bi h depend upon he ea ing en i onmen . Jou nal o Animal Ecology 72: 321–330. 52. Poikonen T, Koskela E, Mappes T, Mills SC (2008) In an icide in he e olu ion o ep oduc i e synch ony: e ec s on ep oduc i e success. E olu ion in p ess. 53. Mills S, G appu o A, Koskela E, Mappes T (2007) Quan i a i e measu e o sexual selec ion wi h espec o he ope a ional sex a io: a compa ison o selec ion indices. P oceedings o he Royal Socie y B: Biological Sciences 274: 143–150. 54. Pa e son S, Lello J (2003) Mixed models: ge ing he bes use o pa asi ological da a. T ends in Pa asi ology 19: 370–375. F equency-Dependen Selec ion PLoS ONE | www.plosone.o g 6 Feb ua y 2008 | Volume 3 | Issue 2 | e1687