Book Read Free

Jim Baen’s Universe

Page 71

by Edited by Eric Flint


  The ge­ne po­ol

  One of the most use­ful con­ven­ti­ons in po­pu­la­ti­on ge­ne­tics is the mo­del of a ge­ne po­ol. We as­su­me that each pa­rent con­t­ri­bu­tes equ­al­ly to a lar­ge po­ol of ga­me­tes. Each of­f­s­p­ring is re­gar­ded as a ran­dom sam­p­le of one egg and one sperm from this po­ol. Each ge­ne is cho­sen ran­domly, as if we we­re dra­wing dif­fe­rent co­lo­ured be­ans from a bag. This over­sim­p­li­fi­ed mo­del bu­il­ding is cal­led “be­an­bag ge­ne­tics” and has a sur­p­ri­singly strong pre­dic­ti­ve po­wer. In 1964, the gre­at Bri­tish ge­ne­ti­cist, J.B.S. Hal­da­ne, wro­te an amu­sing and spi­ri­ted ar­tic­le “A De­fen­ce of Be­an­bag Ge­ne­tics” (Persp. Bi­ol. Med., 7: 343-359).

  Clearly the be­an­bag mo­del is a cru­de rep­re­sen­ta­ti­on of na­tu­re. A re­al po­pu­la­ti­on has in­di­vi­du­als of all ages, so­me dying, so­me cho­osing ma­tes, so­me gi­ving birth, etc, etc. Ke­eping track of all this in­for­ma­ti­on is not only im­p­rac­ti­cal, but of­ten is ne­it­her in­te­res­ting nor im­por­tant. Most qu­es­ti­ons of ge­ne­tic or evo­lu­ti­onary in­te­rest do not re­qu­ire mi­nu­te de­ta­ils. The ge­ne po­ol mo­del will gi­ve us the sa­me kind of in­sight that sim­p­le mo­dels in the physi­cal sci­en­ces do, and of­ten with pre­dic­ti­ons that are suf­fi­ci­ently ac­cu­ra­te for most uses. Fur­t­her it is im­por­tant to be awa­re of that de­par­tu­re from the sim­p­le mo­dels may usu­al­ly be tre­ated with ap­prop­ri­ate mo­di­fi­ca­ti­ons.

  Suppose now that a po­pu­la­ti­on con­sist of in­di­vi­du­als with an ex­t­re­mely long li­fe span. In the tra­di­ti­on of “be­an­bag ge­ne­tics” the po­pu­la­ti­on is ho­moz­y­go­us at the lo­cus de­ter­mi­ning li­fe span; i.e. all ge­not­y­pes are AA and co­de for long-li­fe. Ho­we­ver, in any po­pu­la­ti­on the­re will be so­me ran­dom de­ath ra­te ba­si­cal­ly de­ter­mi­ned from outer so­ur­ces li­ke bi­olo­gi­cal (vi­rus, sic­k­ness, pre­da­ti­on or ac­ci­dents due, for exam­p­le, to na­tu­re ca­tas­t­rop­hes). Thus, rat­her li­ving fo­re­ver, the long-li­ved stra­tegy has to to­le­ra­te the ope­ra­ti­on of a ran­dom de­ath ra­te; that is, the sur­vi­val ra­te, S, is clo­se to 1. This stra­tegy is cal­led evo­lu­ti­onary stab­le if the po­pu­la­ti­on re­j­ects all kinds of in­va­ding al­le­les that re­du­ce the li­fe span. In or­der to in­ves­ti­ga­te this sta­bi­lity, we as­su­me that a small frac­ti­on e of the ge­ne po­ol con­sists of an al­le­le ‘ a’ such that the ge­not­y­pes Aa in­du­ces a shor­ter li­fe span via chan­ges in the birth ra­tes and sur­vi­val ra­tes. We must find the re­qu­ire­ment that the new al­le­le will in­c­re­ase in the ge­ne po­ol.

  A Hardy- Weinberg type mo­del for anal­y­sing the evo­lu­ti­onary sta­bi­lity of in­fi­ni­te long-li­fe

  Since we are only in­te­res­ted in the ge­ne­ral prin­cip­les that go­vern the sta­bi­lity of the long-li­fe stra­tegy we sim­p­lify the po­pu­la­ti­on dyna­mics of the mu­tant al­le­le as fol­lows: On ave­ra­ge they li­ve t ye­ars and rep­ro­du­ces bAa of­f­s­p­ring of which lAa sur­vi­ves the next t ye­ars. The pro­ba­bi­lity of sur­vi­ving t ye­ars for the long-li­ved in­di­vi­du­als is L = St = (S mul­tip­li­ed by it­self t ti­mes). Fi­nal­ly, we ig­no­re the ho­moz­y­go­us mu­tant al­le­le (of ge­not­y­pe aa), sin­ce the­ir abun­dan­ce is of the or­der e(e in the ge­ne po­ol.

  Thus the ini­ti­al ge­not­y­pe fre­qu­en­ci­es are (1-e) of AA and e of Aa whe­re the abun­dant first gro­up of in­di­vi­du­als (AA) ha­ve a very long li­fe span and the se­cond po­ten­ti­al in­va­der gro­up (Aa) only li­ve t ye­ars. In the ge­ne po­ol the fre­qu­ency of A is (1-e)*1 + e*(1/2) = 1-e/2 whi­le the fre­qu­ency of a is q = e-2

  We shall now try to an­s­wer the qu­es­ti­on: Un­der what cir­cum­s­tan­ces is the frac­ti­on of the al­le­le a ex­pec­ted to in­c­re­ase in the ge­ne po­ol?

  In or­der to sol­ve this prob­lem we ne­ed anot­her as­sum­p­ti­on: The po­pu­la­ti­on si­ze is so lar­ge that the ma­ting may be con­si­de­red as a ran­dom mix­tu­re of the ge­nes. We must con­si­der all kind of ma­tings bet­we­en the ge­not­y­pes:

  1. Con­si­der first a long-li­ved fe­ma­le (AA) that ma­tes with a long-li­ved ma­le (AA). Du­ring t ye­ars she pro­du­ces bAA of­f­s­p­ring (of ge­noy­pe AA) that ha­ve a chan­ce of lAA to sur­vi­ve the next t ye­ars. In the po­pu­la­ti­on the pro­ba­bi­lity of such ma­tings is (1-e)*(1-e).

  2. Con­si­der next a long-li­ved fe­ma­le (AA) that ma­tes with a mu­tant short-li­ved ma­le (Aa). Du­ring t ye­ars she pro­du­ces bAA of­f­s­p­ring of which 50% are long-li­ved (AA) with a chan­ce of lAA to sur­vi­ve the next t ye­ars, and 50% are short-li­ved (Aa) with a chan­ce of lAa to sur­vi­ve the next t ye­ars. In the po­pu­la­ti­on the pro­ba­bi­lity of such ma­tings is (1-e)*e.

  3. Then con­si­der a mu­tant short-li­ved fe­ma­le (Aa) that ma­tes with a long-li­ved ma­le (AA). Du­ring t ye­ars she pro­du­ces bAa of­f­s­p­ring of which 50% are long-li­ved (AA) with a chan­ce of lAA to sur­vi­ve the next t ye­ars, and 50% are short-li­ved (Aa) with a chan­ce of lAa to sur­vi­ve the next t ye­ars. In the po­pu­la­ti­on, the pro­ba­bi­lity of such ma­tings is e*(1-e).

  4. Fi­nal­ly con­si­der a mu­tant short-li­ved fe­ma­le (Aa) that ma­tes with a mu­tant short-li­ved ma­le (Aa). Du­ring t ye­ars she pro­du­ces bAa of­f­s­p­ring of which 25% are long-li­ved (AA) with a chan­ce of lAA to sur­vi­ve the next t ye­ars, and 50% are short-li­ved (Aa) with a chan­ce of lAa to sur­vi­ve the next t ye­ars. In the po­pu­la­ti­on the pro­ba­bi­lity of such ma­tings is e*e.

  After t ye­ars the num­ber of long-li­ved and short-li­ved rec­ru­its are res­pec­ti­vely:

  WAA = (1-e)2 bAA lAA + (1-e)*e(1/2)bAA lAA + e*(1-e)(1/2)bAa lAA + e*e(1/4) bAa lAA

  WAa = (1-e)*e(1/2) bAA lAa + e*(1-e)e/2) bAa lAa + e*e(1/2) bAa lAa

  Now, the num­ber of adult long-li­ved in­di­vi­du­als sur­vi­ving a pe­ri­od of t ye­ars will be (1-e)N*S whe­re N is the to­tal po­pu­la­ti­on si­ze. In or­der to ke­ep the po­pu­la­ti­on at a stab­le si­ze of N in­di­vi­du­als, we ha­ve to in­t­ro­du­ce a com­mon de­ath ra­te m for the rec­ru­its of all ge­not­y­pes:

  (1- m)*N* WAA + (1-m)*N* WAa + (1-e)N*L = N

  After t ye­ars the new frac­ti­on of the short-li­ved al­le­le a in the ge­ne po­ol is

  qt =

  Thus we see that the frac­ti­on of the al­le­le a will in­c­re­ase in the ge­ne po­ol if

  ,i.e.

  which is equ­iva­lent to

  ( bAa lAa+(1-e) bAA lAa)» (1- m) WAA+2(1-e) L

  Reordering the terms, fi­nal­ly gi­ves the re­qu­ire­ment that the mu­tant al­le­le co­ding for short-li­ved may in­va­de the ge­ne po­ol:

  Thus, if the mu­tant al­le­le a co­des for a birth ra­te bAa and sur­vi­val ra­te lAa such that this ine­qu­ality is sa­tis­fi­ed it will in­va­de the ge­ne po­ol. But what do­es this ine­qu­ality es­sen­ti­al­ly tell us? In or­der to see the es­sen­ti­al re­qu­ire­ment for in­va­si­on we may use stan­dard tec­h­ni­qu­es in nu­me­ri­cal anal­y­sis. First, sin­ce e is a small qu­an­tity in the or­der of 1100, we may use the stan­dard ap­pro­xi­ma­ti­ons in the fi­nal ex­p­res­si­ons: 1 - e 1 and 1 + e + e2 1 so the ine­qu­ality is prac­ti­cal­ly the sa­me as

  Then we ta­ke a clo­ser lo­ok at the ex­t­ra non-spe­ci­fic mor­ta­lity ra­te m that was in­t­ro­du­ced in or­der to ke­ep the po­pu­la­ti­on at its equ­ilib­ri­um va­lue N. In or­der to un­der­s­tand the mag­ni­tu­de of this qu­an­tity we con­si­der the po­pu­la­ti­on wit­ho­ut the mu­tant al­le­le a (that is all ge­not­y­pes are AA), so the equ­ilib­ri­um equ­ati­on re­du­ces to N* WAA + N* L = N which is se­en to imply:

  Eq
uilibrium con­di­ti­on: WAA + L = 1

  The equ­ilib­ri­um equ­ati­on simply sta­tes that in or­der to sta­bi­li­se the po­pu­la­ti­on, the rec­ru­it­ment to the stock must be ba­lan­ced with the de­ath ra­tes. Now ac­cor­ding to clas­si­cal Dar­wi­ni­an thin­king, suc­ces­sful new mu­tants only ac­com­p­lish small chan­ges. In our ca­se, this me­ans that the new al­le­le a will only slightly mo­dify the rec­ru­it­ment, so the fi­nal adj­us­t­ment of the new short-li­ved rec­ru­its will be small. Hen­ce WAA + WAA + L = 1 so we fi­nal­ly ob­ta­in the fol­lo­wing es­sen­ti­al cri­te­ria for suc­ces­sful in­va­si­on:

  Criteria for suc­ces­sful in­va­si­on of the ge­ne po­ol: lAa» 1

  In words this cri­te­ria simply says that the new al­le­le a will suc­ces­sful­ly in­va­de the ge­ne po­ol if the num­ber of sur­vi­ving mu­tant of­f­s­p­ring to adul­t­ho­od ex­ce­eds 1. No­te that of­f­s­p­ring of ge­not­y­pe Aa is cre­ated in two dif­fe­rent ways: Eit­her the mot­her is of ge­not­y­pe AA and ma­tes with a ma­le of ge­not­y­pe Aa in which ca­se 50% of the of­f­s­p­ring will be Aa, or the mot­her is of ge­not­y­pe Aa in which ca­se al­so 50% of the of­f­s­p­ring will be Aa. This ex­p­la­ins why it is the ave­ra­ge lit­ter si­ze of the two ge­not­y­pes that en­ters the in­va­si­on cri­te­ria. Now the ex­pec­ted num­ber of mu­tant of­f­s­p­ring that re­ac­hes adul­t­ho­od is fo­und by mul­tip­l­ying with the­ir sur­vi­val ra­te lAa. Sin­ce the rec­ru­it­ment of the long-li­ved ge­not­y­pes exactly mat­c­hes the adult mor­ta­lity 1 - L, the to­tal rec­ru­it­ment to the adult gro­up is exactly 1 (i.e. the equ­ilib­ri­um con­di­ti­on). No­te that this is pre­ci­sely whe­re the long-li­fe en­ters the equ­ati­ons: the to­tal rec­ru­it­ment is the rec­ru­it­ment of of­f­s­p­ring that ma­na­ge to li­ve up to adul­t­ho­od PLUS the num­ber of adults sur­vi­ving the pe­ri­od (as men­ti­oned abo­ve, only non-ge­ne­tic ac­ci­dents ca­use a small de­ath ra­te among the long-li­ved in­di­vi­du­als). So the cri­te­ria for suc­ces­sful in­va­si­on of the ge­ne po­ol may be res­ta­ted in words as fol­lows:

  Criteria for suc­ces­sful in­va­si­on of the ge­ne po­ol:

  The rec­ru­it­ment of the short-li­ved in­di­vi­du­als must ex­ce­ed 1

  Note that this is go­od clas­si­cal Dar­wi­nism: The ge­not­y­pe with the hig­hest rec­ru­it­ment will at the end fill up the who­le ge­ne po­ol.

  It is now easy to see why Jim Ba­ens hypot­he­sis is sup­por­ted by this ge­ne­tic mo­del. The who­le is­sue abo­ut why we must die may be for­mu­la­ted the qu­es­ti­on of whet­her a long-li­ved stra­tegy is evo­lu­ti­onary stab­le. A sim­p­le, first ap­pro­xi­ma­ti­on, ge­ne­tic mo­del says that the long-li­ved stra­tegy will only be evo­lu­ti­onary stab­le if it is dif­fi­cult for mu­tant al­ter­na­ti­ves to es­tab­lish a rec­ru­it­ment lar­ger than 1. But as Jim Ba­en has po­in­ted on clas­si­cal bi­olo­gi­cal know­led­ge: the­re are all sorts of dif­fi­cul­ti­es with in­b­re­eding. The­re­fo­re al­most any al­ter­na­ti­ve stra­tegy that re­du­ces energy de­vo­ted to long-li­fe and put it in­to clutch si­ze or vi­ta­lity of of­f­s­p­ring will ha­ve a bet­ter to­tal rec­ru­it­ment than the long-li­ved. The­re­fo­re, the stra­tegy of eter­nal li­fe is evo­lu­ti­onary un­s­tab­le. For hu­man be­ings this has the con­se­qu­en­ce that 125 ye­ars is the ma­xi­mum age.

  Cohort anal­y­sis of the evo­lu­ti­onary sta­bi­lity of in­fi­ni­te long-li­fe

  In this sec­ti­on we de­ri­ve the sa­me re­sult in an al­ter­na­ti­ve mo­del that ta­kes all the ye­ar-clas­ses (co­horts) in­to ac­co­unt.

  Suppose the­re are M adult long-li­ved in­di­vi­du­als con­sis­ting of 1000 ye­ar clas­ses, and let the clutch si­ze per in­di­vi­du­al be bA, the sur­vi­val up to ma­tu­rity is lm and the sur­vi­val ra­te of adults is sA. At equ­ilib­ri­um MbA of­f­s­p­ring are pro­du­ced each ye­ar by 1000 ye­ar clas­ses:

  bA[ MbA­lA + MbA­lA­sA1 + MbA­lA­sA2 + MbA­lA­sA3 + + MbA­lA­sA999] = MbA which is se­en to re­du­ce to

  Equilibrium con­di­ti­on bA­lA = 1 which prac­ti­cal­ly sta­tes that bA­lA = 1 - sA

  i.e. the po­pu­la­ti­on will be in equ­ilib­ri­um if the rec­ru­it­ment to the adult po­pu­la­ti­on (i.e. birth ra­te mul­tip­li­ed by im­ma­tu­re sur­vi­val) ba­lan­ces the adult de­ath ra­te.

  Consider now a po­ten­ti­al in­va­ding al­le­le a that co­des for li­ving only 100 ye­ars as adults, i.e. 10% of the pre­va­iling stra­tegy of 1000 ye­ars. Ini­ti­al­ly the­re are x0 = e M in­di­vi­du­als of the­se new ge­not­y­pes ha­ving a birth ra­te of ba, a sur­vi­vor­s­hip of the im­ma­tu­re pe­ri­od equ­al la and an adult sur­vi­vor­s­hip of sa. With the­se pa­ra­me­ters the ite­ra­ti­on equ­ati­on for the adult mu­tants are

  xl+1 = saxx + ( xl-tm) la = saxl + laxl+1- tm

  Substituting a ge­ne­ral so­lu­ti­on of the form xx = rl x0 gi­ves the cha­rac­te­ris­tic equ­ati­on for the growth of the mu­tant rtm = sartm-1 + la

  In or­der to see whe­re the ro­ot li­es, de­fi­ne the two fun­c­ti­ons f ( r) = rtm and g( r) = sartm-1 + la and no­ti­ce that f (0) = 0 «la = g(0), but sin­ce tm» tm-1, f( r) will even­tu­al­ly be­co­me gre­ater than g( r). Thus so­mew­he­re they in­ter­cept, and the cru­ci­al qu­es­ti­on is whet­her they in­ter­cept to the left of r = 1, in which ca­se the num­ber of in­va­ders will dec­li­ne, or to the right of r = 1, in which ca­se the num­ber of short li­ved in­di­vi­du­als will in­c­re­ase. Sin­ce f (1) = 1 and f (1) = sa + la it fol­lows that the con­di­ti­on for in­ter­cep­ti­on to the right of r = 1, i.e. f(1) «g(1), is 1 «sa + la which fi­nal­ly gi­ves

  Criteria for suc­ces­sful in­va­si­on of the ge­ne po­ol: la» 1 - sa

  In words: The rec­ru­it­ment to the mu­tant short-li­ved in­di­vi­du­als (=ave­ra­ge birth ra­te of long- and short li­ved ge­not­y­pes mul­tip­li­ed with the sur­vi­val pro­ba­bi­lity up to ma­tu­rity) must be lar­ger than the adult de­ath ra­te of the short-li­ved in­di­vi­du­als. No­te that this con­di­ti­on is equ­iva­lent to that fo­und in our pre­vi­o­us Har­dy-We­in­berg type mo­del, so the sa­me dis­cus­si­on ap­pli­es.

  Let us fi­nal­ly see that if the fol­lo­wing ap­pro­xi­ma­ti­on of the po­pu­la­ti­on dyna­mics of the mu­tant xx = ( sa + la )x e M is sub­s­ti­tu­ted in­to the for­mu­la of the Har­dy-We­in­berg mo­del, we ar­ri­ve at the sa­me cri­te­ria for e M mu­tant to in­va­de a ho­moz­y­go­us po­pu­la­ti­on of (1-e) M long-li­ved adults:

  where the se­cond term in the de­no­mi­na­tor is the num­ber of long-li­ved adult that sur­vi­ved the to­tal li­fe span of the short li­ved (i.e. 100 ye­ars) and the third term is the num­ber of adult long-li­ved that rec­ru­its to the adult po­pu­la­ti­on of long li­ved (sin­ce the­se are pre­va­iling we ig­no­re the con­t­ri­bu­ti­on of long-li­ved from the short-li­ved). Using the equ­ilib­ri­um cri­te­ri­on, bA­lA = 1 - sA, we ob­ta­in (1 - e)( sa + la )100 «(l - e) which is equ­iva­lent to sa + la» l i.e. la» l - sa

  Thus, we ar­ri­ve at pre­ci­sely the sa­me ine­qu­ality, de­mon­s­t­ra­ting an in­ner con­sis­tency in our mo­dels.

  A com­pe­ti­ti­on mo­del with tra­de-off bet­we­en rep­ro­duc­ti­on and sur­vi­val.

  While the two pre­vi­o­us mo­dels star­ted with the as­sum­p­ti­on that long-li­fe was an evo­lu­ti­onary stab­le stra­tegy, we now de­ve­lop a mo­del whe­re the dif­fe­rent li­fe his­tory stra­te­gi­es are in di­rect com­pe­ti­ti­on with each ot­her. Ho­we­ver, if a new mu­tant co­de for bet­ter sur­vi­val we shall now ex­p­li­citly re­gard this as a stra­tegy to al­lo­ca­te mo­re energy to sur­vi­val at the cost of energy de­vo­ted to rep­ro­duc­ti­on. The sim­p­lest equ­ati­on for such a tra­de-off mo­
del is bl = b( s) = (1 - s)l whe­re l is a po­si­ti­ve re­al num­ber.

  A first ap­pro­xi­ma­ti­on to the po­pu­la­ti­on growth will then be x1/2+| = [ s + (l - s)l] x1/2

  which has the so­lu­ti­on x1/2 = [ s + (l - s)l]1/2 x0

  From this we see that na­tu­ral se­lec­ti­on will fa­vo­ur a stra­tegy which ma­xi­mi­ses the fun­c­ti­on f ( S) = S + (l - S)l

  Putting the de­ri­va­ti­ve equ­al to ze­ro, f ( S) = l - l (l - S)l-1 = 0, gi­ves the op­ti­mum sur­vi­val ra­te as s* = 1 - () im­p­l­ying an evo­lu­ti­onary stab­le ex­pec­ted li­fe span of T* =

  So al­so in this mo­del will the long-li­ved be se­lec­ted aga­inst be­ca­use they do not pro­vi­de the op­ti­mal com­bi­na­ti­on of birth ra­te and sur­vi­val ra­te. Anot­her con­se­qu­en­ce of tre­men­do­us im­por­tan­ce is that gi­ven a stra­tegy ne­ar the op­ti­mal, the­re will be no se­lec­ti­on aga­inst de­le­te­ri­o­us ge­nes at ages over T*. This al­lows de­ath ge­nes, ra­cing-car ana­logy con­s­t­ruc­ti­on of or­gans, etc. Now hu­man be­ings ha­ve evol­ved from spe­ci­es whe­re the­se mec­ha­nisms al­re­ady ha­ve be­en ope­ra­ting, so this is an evo­lu­ti­onary blind ro­ute with res­pect to long-li­fe un­less it is pos­sib­le to en­ter our ge­nes di­rectly and re­con­s­t­ruct them. We are an ani­mal that has evol­ved over many mil­li­ons of ye­ars to be short li­ved.

  James Pat­rick “Jim” Ba­en (b. 1943) is a no­ted sci­en­ce fic­ti­on pub­lis­her and edi­tor. He star­ted his pub­lis­hing ca­re­er, ap­prop­ri­ately eno­ugh, in the com­p­la­ints de­par­t­ment of Ace Bo­oks af­ter stints in the Army, at CCNY, and in Gre­en­wich Vil­la­ge in the 1960s wor­king as the ma­na­ger of a folk mu­sic cof­fee shop (a “bas­ket ho­use”). Af­ter a ra­pid pro­mo­ti­on to the edi­to­ri­al de­par­t­ment at ACE, he mo­ved to ta­ke Judy-Lynn del Rey’s pla­ce at Ga­laxy Sci­en­ce Fic­ti­on, and suc­ce­eded Ejler Jakob­s­son as edi­tor of Ga­laxy and If in 1974 af­ter a bri­ef trip back to Ace to be as­sis­tant Got­hics edi­tor. Whi­le at Ga­laxy he pub­lis­hed such aut­hors as Jer­ry Po­ur­nel­le, Char­les Shef­fi­eld, Jo­an­na Russ and John Var­ley, and was no­mi­na­ted for se­ve­ral Hu­go Awards. He re­tur­ned to Ace to he­ad the­ir sci­en­ce fic­ti­on li­ne in 1978, wor­king with pub­lis­her Tom Do­herty. When Do­herty left to start Tor in 1980, Ba­en shortly fol­lo­wed and star­ted the SF li­ne the­re. In 1984, he had the op­por­tu­nity to start his own in­de­pen­dent com­pany, Ba­en Bo­oks, now a no­ted SF li­ne. Ba­en Bo­oks has es­tab­lis­hed a lar­ge re­ader­s­hip, pub­lis­hing bo­oks by aut­hors such as Da­vid We­ber, John Rin­go, Eric Flint, Da­vid Dra­ke and Lo­is McMas­ter Bu­j­old.

 

‹ Prev