Tephroseris Integrifolia (Asteraceae)

Investigating genetic factors behind the decline of a threatened plant species – Tephroseris integrifolia (Asteraceae)

Isaksson, Kerstin

2009

Link to publication

Citation for published version (APA): Isaksson, K. (2009). Investigating genetic factors behind the decline of a threatened plant species – Tephroseris integrifolia (Asteraceae). Plant Ecology and Systematics, Lund University.

Total number of authors: 1

General rights Unless other specific re-use rights are stated the following general rights apply: Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal

Read more about Creative commons licenses: https://creativecommons.org/licenses/ Take down policy If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.

LUND UNIVERSITY

PO Box 117 221 00 Lund +46 46-222 00 00 Investigating genetic factors behind the decline of a threatened plant species – Tephroseris integrifolia (Asteraceae)

Investigating genetic factors behind the decline of a threatened plant species – Tephroseris integrifolia (Asteraceae)

Kerstin Isaksson

AKADEMIC DISSERTATION for the degree of Doctor of Philosophy in Plant Ecology and Systematics, to be publicly defended on October 30th 2009 at 10.00 a.m. in Blå hallen at the Department of Ecology, Ecology Building, Sölvegatan 37, Lund, by permission of the Faculty of Sciences at the University of Lund.

The thesis will be defended in English.

Faculty opponent: Professor Jens M. Olesen, University of Aarhus, Denmark.

Lund 2009 A doctoral thesis at a university in Sweden is produced either as a monograph or as a collection of papers. In the latter case, the introductory part constitutes the formal thesis, which summarizes the accompanying papers. These have either already been published or are manuscripts at various stages (in press, submitted or in ms).

Front cover: Tephroseris integrifolia (Fältnocka), photo by Kerstin Isaksson.

Layout: Stig Isaksson Body type: New Century Schoolbook 9·/10.8·

Investigating genetic factors behind the decline of a threatened plant species – Tephroseris integrifolia (Asteraceae) ...... 7

Tack ...... 20

Paper I ...... 25

Paper II ...... 49

Paper III ...... 63

Paper IV ...... 79 This thesis is based on the following papers:

I Isaksson K, Widén B. Inbreeding depression cannot explain the rapid decline of the self-incompatible perennial Tephroseris integrifolia in Sweden. – Submitted.

II Isaksson K. Self-incompatibility does not explain the declining population sizes of Tephroseris integrifolia (Asteraceae). – Manuscript.

III Isaksson K, Nordström S, Widén B. No connection between population size and local genetic variation in AFLP-markers in a threatened plant species (Tephro- seris integrifolia). – Manuscript.

IV Isaksson K, Widén B. Negative association between population size and quantitative genetic variance in Tephroseris integrifolia (Asteraceae) – evidence for a recent bottleneck? – Manuscript.

The participation of the author in the papers:

Paper I: Responsible for planning and carrying out cross-pollinations, for measuring plants in common garden, and for analysing the data. Co-responsible for manuscript.

Paper II: Responsible for planning and carrying out cross-pollinations, for counting seeds, for data analysis and manuscript.

Paper III: Responsible for collecting material in Estonia, for cultivation of plants, for laboratory work, main responsible for data analysis and manuscript.

Paper IV: Responsible for collecting plant material in Britain, Denmark, and Estonia, and for cultivation of plants. Co-responsible for selecting characters to investigate. Main responsible for measuring. Responsible for data analysis. Main responsible for manuscript. Investigating genetic factors behind the decline of a threatened plant species

Investigating genetic factors behind the decline of a threatened plant species – Tephroseris integrifolia (Asteraceae)

Background fragmented as well. Small population Land use change is one of the major size leads to genetic depletion through threats to global species diversity (Pimm genetic drift (Wright 1931), to inbreeding & Raven 2000, Sala et al. 2000, Foley et (Gaggiotti 2003), and to reduced fecund- al. 2005), and in remaining habitat frag- ity as a consequence of a lower supply of ments, species keep going extinct long suitable mates (e.g. Kolb & Lindhorst after fragmentation has taken place 2006, Levin, Kelley, & Starkar 2009). (Pimm & Raven 2000, Foley et al. 2005). Several empirical studies have re- One of the habitat-types which are ported indications of genetic depletion in loosing most species diversity is grass- small and isolated populations – e.g. land (Sala et al. 2000). In Europe, there is often a positive relationship be- nutrient-poor grasslands have been as- tween population size and genetic vari- sociated with human management for ation in molecular marker genes (Lam- thousands of years (Richards 1990), and mi, Siikamäki, & Mustajärvi 1999, they have harboured a lot of species Landergott et al. 2001, Kang, Jiang, & (Pärtel, Bruun, & Sammul 2005), of Huang 2005, Coppi, Mengoni, & Selvi which many are now threatened with ex- 2008), furthermore ¤tness of offspring tinction (Hodgson et al. 2005, Schleuning often increases with distance between et al. 2009). Though most of the loss of parents (Morán-Palma & Snow 1997, grassland cover in Europe took place be- Byers 1998), and lack of self-incompati- fore 1850 (Richards 1990), the more se- bility alleles has been found to reduce vere changes occurred after 1940, seed-set in small populations of self-in- through the introduction of tractors and compatible species (Byers 1995, Wagen- arti¤cial fertilisers (Ratcliffe 1984, ius, Lonsdorf, & Neuhauser 2007). A de- Bernes 1994). However, the general pro- crease of variation in potentially adap- cess towards a more ef¤cient and static tive phenotypic characters with decreas- land use, such as a more rigid division ing population sizes has mostly been between ¤elds and pastures, and the found in experimental populations (Van cessation of rotation of crops started Buskirk & Willi 2006). much earlier than that (Emanuelsson However, in some cases, an unexpec- 2009). tedly high genetic variation has been documented in small populations Genetic depletion in small popula- (Young, Boyle, & Brown 1996, Young, tions Brown, & Zich 1999, Marquardt et al. As habitats have become fragmented, 2007, Williams et al. 2007, Yates et al. plant and animal populations have been 2007, Yoshioka et al. 2007, Gaudel &

7 Investigating genetic factors behind the decline of a threatened plant species

Till-Bottraud 2008), and this is some- cline (and consequently has been put on times explained as the result of gene- national red-lists of threatened species), ¥ow from unknown sources outside the or from a region where it is not rare, and population (Lawrence et al. 2008, not regarded as threatened. Williams et al. 2007), but populations with unexpectedly high genetic variation A threatened grassland species are often also believed to be relicts of Tephroseris integrifolia (L.) Holub (As- larger populations which have suffered teraceae) is a species which grows in nu- from a very recent decrease in popula- trient-poor, calcareous grasslands and tion size, so that not much genetic vari- which in Sweden has suffered a lot from ation has yet been lost (Young, Boyle, & the destruction and fragmentation of Brown 1996, Williams et al. 2007, De these habitats (Regnéll 1976, Widén Almeida Vieira & De Calvalno 2008, 1987). The species has sporophytic self- Luijten et al. 2000, Van Geert, Van Ros- incompatibility (see paper II), it rarely sum, & Triest 2008). Surprisingly, for ¥owers during two successive years, in quantitative characters, variation has the ¤eld ¤rst ¥owering does not occur sometimes been found to be higher in until the fourth year, and it has no seed- small populations than in large popula- bank (Widén 1987). This species has con- tions (e.g. Widén & Andersson 1993, tinued to decrease in population size and Waldmann 2001) and in several bottle- number even after fragmentation has neck experiments, quantitative vari- ceased and in spite of the implementa- ation has been found to increase during tion of a number of conservation the ¤rst generations (Van Buskirk & measures (Widén & Wetterin 1999). Of Willi 2006). The most plausible explana- totally 42 populations identi¤ed since tion for a negative relationship between the 19th century, only six remain today, population size and variation in quanti- and they are still decreasing in size (see tative characters is that during the ¤rst paper I). In this thesis, I examine if the generations of inbreeding, variation in decline in Swedish T. integrifolia has had quantitative characters can sometimes any effect on local genetic variation in increase as the result of an increased the remaining populations, and discuss homozygosity for previously rare alleles, whether genetic depletion might be a leading to the expression of novel pheno- contributory cause of the continuing de- types (Willis & Orr 1993). cline of the species. According to a meta-study by Spiel- man, Brook, & Frankham (2004) and a Methods review by Frankham (2005) species do In this thesis, different aspects of local not go extinct before they are impacted genetic variation are studied, with focus by genetic factors. E.g. 77% of 170 inves- on the declining Swedish T. integrifolia tigated taxa listed as threatened by the populations, but also including Estonian IUCN (i.e. globally red-listed species) (papers III and IV), British (paper IV), had lower heterozygosity than non- and Danish (paper IV) populations. In threatened relatives (Spielman, Brook, all, 23 contemporary, North European & Frankham 2004). populations are included in these If there is a difference in genetic vari- studies, together with herbarium data ation between globally red-listed and and notations from about 30 historical non red-listed species, we ought also to Swedish populations. be able to detect differences in genetic All of the studied populations seem to variation in different populations of the be dependent on grazing or other hu- same species, depending on the size of man-induced activities – with the excep- the populations, and depending on tion of one Danish population, Hamborg, whether we sample from a region where where the plants grew on a steep slope the species is suffering from severe de- where erosion minimises competition

8 Investigating genetic factors behind the decline of a threatened plant species

(Isaksson pers. obs.). It might also be included in the ¤rst study in this thesis, possible that at least part of one Est- to give a more detailed picture of popula- onian population, Vohilad does not de- tion ¥uctuations and extinctions. pend on grazing for its persistance – we found most of the plants on a beach with Inbreeding experiment (paper I) chalk gravel, suffering relatively little The main focus of the ¤rst study is an in- competition from other plants, but it is breeding experiment on plants culti- reasonable to believe that the species vated from seeds collected in three of the had been more common further inland Swedish populations, Benestad, Grödby, when the grasslands were still grazed and Råby. Plants were crossed in dif- (Isaksson pers. obs.). ferent categories, depending on where In Sweden (Gärdenfors 2005) and their ¤eld mothers had grown in the Britain (Chef¤ngs & Farrell 2005), T. in- populations – between-sib cross (i.e. be- tegrifolia is listed as endangered ac- tween plants that had the same ¤eld cording to the IUCN classi¤cation. In mother), within patch cross (i.e. between Denmark it is regarded as rare (Ingelög plants whose ¤eld mothers had grown et al. 1993), but it is not present on the less than 10 metres away from each Danish national red-list (Den danske other), cross between patches (i.e. be- rødliste 2004), and Ingelög et al. (1993) tween plants whose ¤eld mothers had report it as present, but not threatened, grown between 25 and 400 metres away in Estonia. from each other, but within the same In all of our experimental studies we population), and in some cases cross be- investigated local genetic variation tween populations. We studied the sur- (sometimes within 10 metres distance vival and ¥owering rates of the offspring, and sometimes within a 10 m2 area) to and we measured characters typically re- make it possible to compare populations lated to ¤tness in plants. of different sizes. Self-incompatibility experiment (paper II) The history of Swedish populations from In the second study of this thesis, I made 1820 till present times (paper I) use of the cross-pollinations performed A survey of historical records of the spe- to produce the plants in the ¤rst study. cies together with measures of herbar- Since T. integrifolia is self-incompatible, ium specimens in the Botanical Museum it is not only unable to produce seeds at Lund University were performed to after self-pollination, but pollinations get an estimate of when the main extinc- between different plants which are gen- tions of the Swedish populations oc- etically similar for the self-incompati- curred, and whether it is possible to ob- bility locus (S-locus) do not produce any serve any signs of inbreeding depression seeds either. This is a system which de- just prior to population extinctions, such creases the risk of inbreeding, since close as decreased plant sizes or a decreased relatives are often incapable of pro- number of ¥ower heads. ducing offspring together (De Nettan- court 2001, Castric & Vekemans 2004). If A 30-year census of eight Swedish popu- the number of S-alleles within a popula- lations (paper I) tion is very low, the chances of seed-set The eight Swedish populations remain- decreases for individual plants (De- ing in 1980 have been studied more or mauro 1993). The purpose was to evalu- less thoroughly since this time (e.g. Wi- ate whether, and to what extent, the oc- dén 1987, 1991, 1993, Widén & Anders- casionally observed low seed-set in small son 1993, Widén & Wetterin 1999) and patches of this species (Widén 1993) previously unpublished data on number could be related to a decreased genetic of ¥owering plants, regeneration, and variation at the S-locus. deaths of seedlings and adult plants are

9 Investigating genetic factors behind the decline of a threatened plant species

Variation in AFLP markers (paper III) ish populations remaining after 1980 In the third study we investigated local showed strong ¥uctuations in number of (within ~10 m2) genetic variation in ¥owering plants between years, with AFLP markers, to assess the relation- minima after dry summers (paper I ship between population size and local Fig.5). During 1980–2009 two popula- genetic variation, and to determine tions went extinct, and another two are whether local genetic variation differed now close to extinction. After 1990, the between Swedish (i.e. red-listed) and number of ¥owering plants has ¥uctu- Estonian (not red-listed) populations. ated between much lower values than Two Swedish and nine Estonian popula- during the previous years. tions were used. Inbreeding: The results of the inbreeding Quantitative variation (paper IV) experiments (paper I) varied between In the fourth study we investigated the different kinds of traits. I will here make local quantitative variation in thirteen a short summary of the effects of in- North European T. integrifolia popula- breeding on seed germination, survival, tions – three from Britain, where it is ¥owering, and on morphological traits: red-listed (Chef¤ngs & Farrell 2005), three from Denmark, where it is rare Seed germination (inbreeding): For the (Ingelög et al. 1993), but not red-listed Benestad and Grödby populations, there (Den danske rødliste 2004), three from were signi¤cant differences between Sweden, where it is red-listed (Gärden- crossing categories for most traits re- fors 2005), and four from Estonia, where lated to seed germination (paper I it is neither red-listed nor rare (Ingelög Fig.7a–c). Cross-pollinations between et al. 1993). We use the term broad sense plants from different patches resulted in heritability for the phenotypic express- seeds with higher germination rate and ion of genetic variation. shorter germination times than within patch cross-pollinations, as well as a Results higher survival rate for seeds that had Historical populations: The survey of been stored for 1.5 years, while seeds re- historical Swedish populations showed sulting from sib crosses had the lowest that the most drastic decrease in popula- germination rates, the longest germination number during the investigated tion times, and the lowest survival rate period probably occurred between 1880 after storage. For the Råby population, and 1900 (paper I Fig.2–3). Of 42 identi- there were hardly any differences in seed ¤ed populations, 15 remained at the turn germination between crossing cat- of the 20th century, and only six at the egories, although the trends seemed to turn of the 21st century. There are no in- be the same. However, the Råby seeds dications of populations suffering from had generally much lower germination inbreeding depression before they went rates than the Benestad and Grödby extinct – rather, plants seem to have seeds regardless of crossing category. grown taller with time (paper I Fig.4), while the number of ¥ower heads has Flowering and survival (inbreeding): For been relatively constant. This could be a survival and ¥owering, both for plants result of habitats being overgrown by transferred to a common garden and for bushes before they were destroyed – plants transplanted to the site of an ex- since growing taller as a response to tinct population (paper I Fig.8–9) we light shortage has been documented for found more signi¤cant differences be- T. integrifolia (Widén & Andersson tween crossing categories for Grödby and 1993). Råby plants than for Benestad plants – generally, the further away the ¤eld Census: The 30-year census of the Swed- parents had grown in the natural popu-

10 Investigating genetic factors behind the decline of a threatened plant species

lations, the higher the ¤tness of the the rest, rather than between within offspring. patch and between patch crosses.

Quantitative characters (inbreeding): For Molecular variation: The comparison of quantitative characters, we found no sig- local genetic variation in AFLP markers ni¤cant differences between crossing- (paper III) in eleven populations (two categories, with the exception of height, Swedish and nine Estonian) indicated no for which we found signi¤cant differ- difference in variation depending on ences for the Benestad plants – with be- population size, and no difference between-patch crosses resulting in the tween Swedish and Estonian popula- tallest offspring and sib-crosses in the tions (paper III Fig.1). However, the shortest (paper I Fig.10). There were no Swedish populations were genetically signi¤cant differences between crossing different from the Estonian populations, categories for Grödby or Råby. However, and one isolated population from an Est- the Benestad plants in cultivation were onian island was genetically different generally taller than the plants from the from the rest of the Estonian populations other populations, regardless of cros- (paper III Fig.2a–c). sing-category. On the other hand, the Be- nestad plants were signi¤cantly shorter Quantitative variation: In the fourth than the Grödby plants in the natural study (paper IV), we found that heri- populations. tabilities were higher for the Swedish populations than for the rest, with the Self-incompatibility: In natural popula- Estonian populations having the lowest tions, there was a clear relationship be- heritabilities (paper IV Fig.3). Of 18 tween number of ¥owering stems in a characters, the Swedish and the Danish patch and seed-set (paper I Table 1, populations had signi¤cant heritabilities Fig.6). Benestad generally had smaller for 14, the British for ten, and the Est- patches, and consequently lower seed- onian for seven characters. set. After cross-pollination in the green- house, the relationship between cros- Discussion sing-category and seed-set was the The results of the inbreeding experiment strongest for the Benestad plants (paper (paper I), the self-incompatibility experi- II Fig.4–5). However, for plants that set ment (paper II), and the AFLP analysis seeds, seed-set was higher for the Bene- (paper III) are relatively consistent with stad plants than for the others, except for one another. There is no evidence for cross-pollinations between sibs (paper II genetic depletion in the small Swedish Fig.6). For all three populations, more populations (papers I, II, and III). Most than 60% of the cross-pollinations be- differences in ¤tness for different levels tween sibs gave a seed-set above 2%, and of inbreeding are between sib-crosses for within-patch cross-pollination ~80% and the rest, rather than between within of the Benestad plants and ~90% of the patch and between patch crosses (papers Grödby and Råby plants had a seed-set I and II). There is no relationship be- above 2%, which was my de¤nition of a tween population size and local genetic compatible cross, i.e. a cross between in- variation in eleven investigated popula- dividuals which are genetically different tions (paper III) and there is no dif- at the S-locus. ference between Swedish and Estonian populations in local genetic variation Inbreeding and self-incompatibility – (paper III). concluding remark: Generally, both for However, in the fourth study (paper the inbreeding experiment and the self- IV), we found a negative relationship be- incompatibility study, the strongest dif- tween population size and the level of ferences were between sib-crosses and variation in quantitative characters. And

11 Investigating genetic factors behind the decline of a threatened plant species

the Swedish populations had signi¤cant ever, they do not present any theories heritabilities for more characters than about possible differences in the import- the Estonian populations. A possible ex- ance of genetic factors for different spe- planation of these ‘reverse’ results is that cies extinctions. the Swedish populations are in the early stages of a bottleneck, since an initial ef- Why is there so little effect of fragmenta- fect of decreased population size could be tion on genetic variation in T. integrifolia an increase in heritability as a result of populations? homozygosity for previously rare reces- Is there a difference between extinction sive alleles, which might become more events driven by genetic factors, and ex- frequent through genetic drift (Willis & tinction events where genetic factors are Orr 1993). This phenomenon has been negligible? If, for instance, we compare observed in several empirical studies, ac- my study of self-incompatibility in T. in- cording to a meta-study by Van Buskirk tegrifolia with that performed by Wagen- & Willi (2006). ius, Lonsdorf, & Neuhauser (2007) on Given the results of the studies pre- the North American species Echinacea sented in this thesis, it is not likely that angustifolia, we can conclude that the genetic depletion has played any im- two species have many similarities. E. portant part in the decrease of T. in- angustifolia (Asteraceae) is a grass-land tegrifolia in Sweden during the past de- species with sporophytic self-incompati- cades. And given how rapidly the popu- bility, longevity is about 16–44 years lations are decreasing in size, it is likely (compared with 12–58 years for T. in- that they will reach extinction before tegrifolia (Widén 1987)), ¤rst ¥owering they are impacted by genetic factors. In occurs after 2–9 years, it does not ¥ower their study on genetic variation in every year, and fragmentation of its na- threatened and non-threatened species, tural habitats started between 1870 and Spielman, Brook, & Frankham (2004) 1950 (Wagenius, Lonsdorf, & Neuhauser compared closely related taxa in dif- 2007). Their results indicate that the low ferent animal and plant groups. They seed-set in fragments of E. angustifolia found that 77% of the studied red-listed populations is due to the loss of S-alleles species had less genetic variation than through drift, while in T. integrifolia, related common species. That decrease though self-incompatibility seems to in population size leads to genetic de- play a part in the lower seed-set in Bene- pletion is further con¤rmed in a meta- stad, most within patch cross-pollina- analysis by Leimu et al. (2006), which tions gave at least some seed-set, indi- covers 41 species and also by studies of cating high local genetic variation. Are single species (e.g. Lammi, Siikamäki, & there any factors – intrinsic or extrinsic – Mustajärvi 1999, Landergott et al. 2001, which make different species react dif- Kang, Jiang, & Huang 2005, Coppi, ferently to decreasing population sizes, Mengoni, & Selvi 2008). In the four or is it just a matter of time (as Spiel- studies presented in my thesis, popula- man, Brook, & Frankham (2004) sug- tions of only one species are studied in gest) before hitherto unaffected species regions where it is red-listed and in re- are genetically affected by their rare- gions where it is not red-listed. ness? Spielman, Brook, & Frankham (2004) hypothesise that of the 23% of the Changing dispersal mechanisms and studied rare species which had not lower ‘colonization de¤cit’ heterozygosity than closely related non- Several plant species which have tradi- rare species, many will be genetically tionally been characterised as wind-dis- affected by the decreased population persed, among them Asteraceae species sizes before they go extinct, which might with achenes with pappi, have been speed up the extinction process. How- found to been epizoochorous, i.e. dis-

12 Investigating genetic factors behind the decline of a threatened plant species

persed in the fur of animals (Courvreur lated for a certain time-period, rather et al. 2004, Römermann, Tackenberg, & than a certain number of generations, Poschlod 2005, De Pablos & Peco 2007). and that, as a result, extinction risks of Experimental studies of putatively wind- long-lived plants and animals are se- dispersed seeds have also showed that verely underestimated. It would, they the achenes of some Senecio species also conclude, be easier to predict the ex- (closely related to Tephroseris and with tinction risk of a taxon if it is calculated similar pappi) only disperse a few metres for a certain number of generations, at normal wind-speeds (Sheldon & rather than a certain time-period Burrows 1973). Apart from the destruc- (O’Grady et al. 2008). It might be added tion and fragmentation of semi-natural though, that for long-lived species, since grasslands in Europe during the past the extinction process is slower, we have centuries, isolation seems to have been more time to rescue them. further enhanced as livestock became The average life-length of T. integri- more immobile, being enclosed within folia plants is approximately 40 years, smaller pieces of grass-land, and not given the calculations of Widén (1987), moving so much between pastures (e.g. though it varies between populations. Bruun & Fritzbøger 2002). As pointed About 120 years have passed since most out by Wessels et al. (2008) and Ozinga of the Swedish T. integrifolia populations et al. (2009), red-listed or declining spe- disappeared (paper I) – a time-period cies have been found to be over-repre- during which the plants in a T. integri- sented among epizoochorous plants, and folia population would be replaced three Ozinga et al. (2009) identify what they times. The introduction of tractors and call a ‘colonization de¤cit’ as responsible arti¤cial fertilisers – which implied a for the decline of many grassland spe- radical change of the remaining semi-na- cies. In paper III, we hypothesise that tural grasslands – started about 70 years the land use changes which have led to ago (Ratcliffe 1984, Bernes 1994), slight- the extinctions of T. integrifolia popula- ly less than the time it would take for the tions have also disrupted the dispersal plants in a population to be replaced dynamics of this species – even if move- twice. According to the census described ments of cattle and sheep might have in paper I, the number of ¥owering T. been modest in southern Sweden com- integrifolia plants has decreased to pared with areas where transhumance about one sixth during the past 30 years, (seasonal movement of live-stock) was i.e. in less than the average life-length of widely practiced (e.g. described by an established plant. It is therefore prob- Fischer, Poschlod, & Beinlish 1996). ably not too bold to assume that T. integrifolia in Sweden is one of the species What will happen to T. integrifolia which might be driven to extinction be- in Sweden? fore it is affected by genetic factors (cf. In a meta-study of 100 vertebrate taxa, Spielman, Brook, & Frankham (2004)). O’Grady et al. (2008) ¤nd a strong negative correlation between generation What is the conservation value of T. in- length and the minimum population size tegrifolia in Sweden? which gives a 90% chance of population The conservation value of a species is persistence, but no relationship between often estimated on a national basis – a generation length and the minimum species being rare in a state gives it a population size that gives a 90% chance high conservation value there, regard- of persistence for a ¤xed number of gen- less of the species’ status in neighbour- erations. They conclude that the reason ing countries (Lesica & Allendorf 1995). why the results of estimations of extinc- The result is that conservation efforts tion risks varies so much between taxa is are often spent on species in the periph- that extinction risks are usually calcu- ery of their distribution, where the gen-

13 Investigating genetic factors behind the decline of a threatened plant species

etic variation is the lowest, rather than implemented very soon, or it will be too in the centre (Lesica & Allendorf 1995). late. The Swedish T. integrifolia populations are certainly peripheral compared to the Fragmentation and population dy- overall distribution (cf. Smith 1979). namics – alternative scenarios What measures would make the popula- In all of the studies in this thesis, local tions increase? How much would it cost? genetic variation was investigated, to And would it be worth the effort? If, as I make it possible to compare populations suggest, T. integrifolia is one of the spe- of different sizes, as well as to ¤nd indi- cies which suffer from ‘colonization de- cations of loss of genetic variation after ¤cit’ (cf. Ozinga et al. 2009), then it might population sizes have decreased. What be favoured by activities which lead to an does local genetic variation tell us about enhanced transportation of seeds, e.g. the condition of a population? And what the movement of hay or animals between does it tell us about the possibilities for fragments of semi-natural grassland. It populations to recover, should they ever is likely that such activities might favour get the chance to increase in size? Let us other grassland species as well, some- assume three plant populations, A, B, thing which would increase the value of and C, of equal sizes, densities, dispersal such measures. However, if there is some dynamics, and mortalities. The number method through which T. integrifolia of new plants which can be established could be rescued in Sweden, it must be each generation is density dependent:

Populations A and B go through a fragmentation process, while population C remains large and thriving:

In population A, due to some additional habitat fragments, but no dispersal oc- environmental changes, all establish- curs between fragments. Consequently, ment of new individuals ceases, but mor- plant density will remain the same with- tality of already established individuals in the fragments in population B, while is the same, while in population B, the density in population A will decrease as plants continue to reproduce, and off- plants die, and no new seedlings are es- spring can be established within the tablished:

14 Investigating genetic factors behind the decline of a threatened plant species

After a few years, half of the plants pre- area corresponding to the fragment-sizes sent when habitats A and B were frag- of populations A and B. This corresponds mented have died. Twenty random indi- to the analyses of local genetic variation viduals from each population are described in this thesis. sampled for genetic analysis within an

In population A, all of the sampled indi- clusion that population A is ’healthier’ viduals are old – having been success- than population B, and possibly more fully established before the habitat was valuable for conservation. For neither fragmented – while in populations B and population A nor population B has loss of C half of the sampled individuals are old genetic variation contributed to the de- and half are young. In population B, all crease in population sizes. The local gen- of the younger plants are the offspring of etic variation might not have changed the older plants within the fragment, much since fragmentation – however, all while in population C, parents of young the genetic variation outside the frag- plants might be found both within and ments was lost during the fragmentation outside the sampled area. If we know process. nothing about the change of dynamics in Let us assume that the habitats of populations A and B, and if we do not populations A and B are restored, so that compare the plant densities of the three new plants can again be established both populations, we might draw the con- within and between the fragments:

15 Investigating genetic factors behind the decline of a threatened plant species

After another few years, a new sample is but in populations A and B from an area taken from each of the populations of the outside the fragments that were pre- same size as at the previous sampling, served during the fragmentation:

How much genetic variation will we be rapid decline in T. integrifolia was not able to detect? And how will the three caused by genetic factors. populations differ? How much genetic variation is de- References tected at the two sampling events in the Bernes C.1994. Biologisk mångfald i Sverige: three populations A, B, and C, will de- en landstudie. Monitor 14, Statens natur- pend very much on genetic structure and vårdsverk, Solna. population dynamics prior to the frag- Bruun HH, Fritzbøger B. 2002. The Past mentation events. Leblos Estoup, & Impact of Livestock Husbandry of Plant Seeds in the Landscape of Denmark. Ambio Streiff (2006) demonstrated in their 31:425–431. simulations where different sized popu- Byers DL. 1995. Pollen Quantity and Quality lations suffered an instantaneous loss of as Explanations for Low Seed Set in Small individuals and habitat area, that local Populations Exempli¤ed by Eupatorium genetic variation decreases more quickly (Asteraceae). American Journal of Botany in panmictic populations than in popula- 82:1000–1006. tions with strong isolation by distance, Byers DL. 1998. Effect of cross proximity and and that this effect was stronger for very progeny ¤tness in a rare and a common species of Eupatorium (Asteraceae). American small sampling areas. Journal of Botany 85:644–653. The Swedish T. integrifolia popula- Castric V, Vekemans X. 2004. Plant self-in- tions might, after the initial habitat de- compatibility in natural populations: a structions during the late 19th century, critical assessment of recent theoretical have encountered a situation similar to and empirical advances. Molecular Ecology what happened in population B above, 13:2873–2889. managing to reproduce for a few gener- Chef¤ngs CM, Farrell L. (eds), Dines TD, ations, but with lower effective popula- Jones RA, Leach SJ, McKean DR, Pearman DA, Preston CD, Rumsey FJ, Taylor I. tion sizes than before, which would ex- 2005. The Vascular Plant Red Data List for plain the greater variation in quantitat- Great Britain. Species Status 7:1–116. Joint ive characters in the small Swedish Nature Conservation Committee, Peter- populations than in the large Estonian borough. populations, indicating an early bottle- Coppi A, Mengoni A, Selvi F. 2008. AFLP neck (paper IV). However, since the ¤ngerprinting of Anchusa (Boraginaceae) 1980s the Swedish T. integrifolia popula- in the Corso-Sardinian system: Genetic di- tions seem to have been in a situation versity, population differentiation and conservation priorities in an insular endemic more similar to that of population A, group threatened with extinction. Biologi- with almost no successful reproduction. cal Conservation 141:2000–2011. We do not know why this happened, but Couvreur M, Christiaen B, Verheyen K, based on the studies presented in this Hermy M. 2004. Large herbivores as mobile thesis, we can assume that the recent links between isolated nature reserves

16 Investigating genetic factors behind the decline of a threatened plant species

through adhesive seed dispersal. Applied ¥ora of Central England: processes and Vegetation Science 7:229–236. prospects. Basic and Applied Ecology De Almeida Vieira F, De Calvalho D. 2008. 6:107–118. Genetic structure of an insect-pollinated Ingelög T, Andersson R, Thernborg M. 1993 and bird-dispersed tropical tree in vegeta- (eds.) Red Data Book of the Baltic Region. tion fragments and corridors: implications Part 1. Lists of threatened vascular plants for conservation. Biodivers. conserv. 17: and vertebrates. Swedish Threatened Spe- 2305–2321. cies Unit, Uppsala, in cooperation with the De Nettancourt D. 2001. Incompatibility and Institute of Biologya, Riga. Incongruity in Wild and Cultivated Plants. Kang M, Jiang M, Huang H. 2005. Genetic Second, totally revised and enlarged edi- Diversity in Fragmented Populations of tion. Springer-Verlag. Berchemiella wilsonii var. pubipetiolata De Pablos I, Peco B. 2007. Diaspore morphol- (Rhamnaceae). Annals of Botany 95:1145– ogy and the potential for attachment to ani- 1151. mal coats in Mediterranean species: an ex- Kolb A, Lindhorst S. 2006. Forest fragmenta- periment with sheep and cattle coats. Seed tion and plant reproductive success: a case Science Research 17:109–114. study in four perennial herbs. Plant Ecol- Demauro MM. 1993. Relationship of Breeding ogy 185:209–220. System to Rarity in the Lakeside Daisy Lammi A, Siikamäki P, Mustajärvi K. 1999. (Hymenoxys acualis var. glabra). Conserva- Genetic Diversity, Population Size, and Fit- tion Biology 7:542–550. ness in Central and Peripheral Populations Den danske rødliste 2004. Den danske rød- of a Rare Plant Lychnis viscaria. Conserva- liste. Fagdatacenter for Biodiversitet og tion Biology 13:1069–1078. Terrestrisk Natur (B-FDC). – Danmarks Landergott U, Holderegger R, Kozlowski G, Miljøundersøgelser. Found at: http:// Schneller, JJ. 2001. Historical bottlenecks redlist.dmu.dk. decrease genetic diversity in natural popu- Emanuelsson U. 2009. Europeiska kultur- lations of Dryopteris cristata. Heredity landskap. Hur människan format Europas 87:344–355. natur. Forskningsrådet Formas. Lawrence HA, Taylor GA, Millar CD, Fischer SF, Poschlod P, Beinlich B. 1996. Ex- Lambert DM. 2008. High mitochondrial perimental Studies on the Dispersal of and nuclear genetic diversity in one of the Plants and Animals on Sheep in Calcareous world’s most endangered seabirds, the Grasslands. The Journal of Applied Ecology Catham Island Daiko (Pterodroma mag- 33:1206–1222. entae). Conservation genetics 9:1293–1301. Foley JA, DeFries R, Asner GP, Barford C, Leblos R, Estoup A, Streiff R. 2006. Genetics Bonan G, Carpenter SR, Chapin FS, Coe of recent habitat contraction and reduction MT, Daily GC, Gibbs HK, Helkowski JH, in population size: does isolation by dis- Holloway T, Howard EA, Kucharik CJ, tance matter? Molecular Ecology 15:3601– Monfreda C, Patz JA, Prentice IC, 3615. Ramankutty N, Snyder PK. 2005. Global Leimu R, Mutikainen P, Koricheva J, Fischer Consequences of Land Use. Science 309: M. 2006. How general are positive relation- 571–574. ships between plant population size, ¤t- Frankham R. 2005. Genetics and extinction. ness, and genetic variation? Journal of Biological conservation 126:131–140. Ecology 94: 942–952. Gaggiotti OE. 2003. Genetic threats to popu- Lesica P, Allendorf FW. 1995. When Are lation persistence. Ann. Zool. Fennici Peripheral Populations Valuable for Con- 40:155–168. servation? Conservation Biology. 9:735– Gärdenfors U. (ed.) 2005. Rödlistade arter i 760. Sverige 2005 – The 2005 Red List of Swed- Levin DA, Kelley CD, Starkar S. 2009. En- ish Species. ArtDatabanken, SLU, Uppsala. hancement of Allee effects in plants due to Gaudel M, Till-Bottraud I. 2008. Genetic self-incompatibility alleles. Journal of Ecol- structure of the endangered perennial plant ogy 97:518–527. Eryngium alpinum (Apiaceae) in an alpine Luijten SH, Dierick A, Gerard J, Oostermeijer valley. Biological Journal of the Linnean B, Raijmann LEL, Den Nijs HCM. 2000. Society 93:667–677. Population Size, Genetic Variance and Re- Hodgson JG, Grime JP, Wilson PJ, Thompson productive Success in a Rapidly Declining, K, Band SR. 2005. The impacts of agricul- Self-Incompatible Perennial (Arnica mon- tural change (1963–2003) on the grassland tana) in the Netherlands. Conservation

17 Investigating genetic factors behind the decline of a threatened plant species

Biology 14:1776-1787. Oesterheld M, Poff NL, Sykes MT, Walder Marquardt PE, Echt CS, Epperson BK, BH, Walker M, Wall DH. 2000. Global Pubanz DM. 2007. Genetic structure, diver- Biodiversyt Scenarios for the Year 2100. sity and inbreeding of eastern white pine Science 287:1770–1774. under different management conditions. Schleuning M, Niggemann M, Becker U, Canadian Journal of Forest Research Matthies D. 2009. Negative effects of habi- 37:2652–2662. tat degradation and fragmentation on the Morán-Palma P, Snow AA. 1997. The effect of declining grassland plant Trifolium monta- interplant distance in mating success in num. Basic and Applied Ecology 10:61–69. federally threatened, self-incompatible Hy- Sheldon JC, Burrows FM. 1973. The Dis- menoxys herbacea = H. aucualis var. glabra persal Effectiveness of the Achene – Pappus (Asteraceae). American Journal of Botany Units of Selected Compositae in Steady 84:233–238. Winds with Convection. New Phytologist O’Grady, J. J., Reed, D. H., Brook, B. W., & 72:665–675. Frankham, R. 2008. Extinction risk scales Smith UK. 1979. Senecio integrifolius (L) better to generations than to years. Animal Clairv. (Senecio campestris (Retz.) DC.). Conservation. 11:442–451. Journal of Ecology 67:1109–1124. Ozinga W, Römermann C, Bekker RM, Spielman D, Brook BW, Frankham R. 2004. Prinzing A, Tamis WLM, Schaminée JHJ, Most species are not driven to extinction be- Henneskens SM, Thompson K, Poschlod P, fore genetic factors impact them. PNAS Kleyer M, Bakker JP, Van Groenendael JM. 101:15261–15264. 2009. Dispersal failure contributes to plant Van Buskirk J, Willi Y. 2006. The Change in losses in NW Europe. Ecology Letters Quantitative Genetic Variance with In- 12:66–74. breeding. Evolution 60:2428–2434. Pärtel M, Bruun HH, Sammul M. 2005. Biodi- Van Geert A, Van Rossum F, Triest L. 2008. versity in temperate European grasslands: Genetic diversity in adult and seedling origin and conservation. Integrating ef- populations of Primula vulgaris in a frag- ¤cient grassland Farming and Biodiversity. mented agricultural landscape. Conserva- Proceedings of the 13th International Occa- tion Genetics 9:845–853. sional Symposium of the European Grass- Wagenius S, Lonsdorf E, Neuhauser C.2007. land Federation Tartu, Estonia 29-31 Patch Aging and the S-Allee Effect: August 2005. (Eds. R. Lillak, R. Viirtalt, A. Breeding System Effects on the Demo- Linke, & V. Geherman). graphic Response of Plants to Habitat Frag- Pimm SL, Raven P. 2000. Extinction by num- mentation. The American Naturalist bers. Nature 403:843–845. 169:383–397. Ratcliffe DA. 1984. Post-Medieval and Recent Waldmann P. 2001. Additive and non-additive Changes in British Vegetation: the Culmi- genetic architecture of two different-sized nation of Human In¥uence. New Phytol. populations of Scabiosa canescens. Heredity 98:73–100. 86:648–657. Regnéll G. 1976. Naturreservatet Benestads Wessels S, Eichberg C, Storm C, Schwabe A, backar – vegetation och ¥ora. Svensk Bota- 2008. Do plant-community-based grazing nisk Tidskrift 70: 17–42. regimes lead to epizoochorous dispersal of Richards JF. 1990. Land Transformation. In high proportions of target species? Flora The Earth As Transformed by Human 203:304–326. Action. Global and Regional Changes in the Widén B. 1987. Population biology of Senecio Biosphere over the Past 300 Years. Ed. B. L. integrifolius (Compositae), a rare plant in Turner II) pp 164-178. Cambridge Univer- Sweden. Nordic Journal of Botany 7: 686– sity Press with Clark University. Cam- 704. bridge. Widén B. 1991. Phenotypic selection on Römermann C, Tackenberg O, Poschlod P. ¥owering phenology in Senecio integri- 2005. How to predict attachment potential folius, a perennial herb. Oikos 61: 205–215. of seeds to sheep and cattle coat from Widén B. 1993. Demographic and genetic ef- simple morphological seed traits. Oikos fects on reproduction as related to popula- 110:219–230. tion size in a rare, perennial herb, Senecio Sala OE, Chapin FS III, Armest JJ, Berlow E, integrifolius (Asteraceae). Biological Jour- Bloom¤eld J, Dirzo R, Huber-Sanwald E, nal of the Linnean Society 50:179–195. Huenneke LF, Jackson RB, Kinzig A, Widén B, Andersson S. 1993. Quantitative Leemans R, Lodge DM, Mooney HA, genetics of life-history and morphology in a

18 Investigating genetic factors behind the decline of a threatened plant species

rare plant, Senecio integrifolius. Heridity Composition of the pollinator community, 70:503–514. pollination and the mating system for a shrub Widén B, Wetterin M. 1999. Åtgärdsprogram in fragments of species rich kwogan in south- för bevarande av fältnocka (Tephroseris in- west Western Australia. Biodiversity and tegrifolia). Stockholm: Naturvårdsverket. Conservation 16:1379–1395. Williams DA, Wang Y, Borchetta M, Gaines Yoshioka Y, Hojo M, Iwata H, Ninomiya S, MS.2007. Genetic diversity and spatial Ohsawa R. 2007. Pattern of geographical structure of a keystone species in frag- variation in petal shape in wild populations of mented pine rockland habitat. Biological Primula sieboldii E. Morren. Plant Species Conservation 138:256–268. Biology 22:87–93. Willis JH, Orr HA. 1993. Increased heritable Young A, Boyle T, Brown T. 1996. The popula- variation following population bottlenecks: tion genetic consequence of habitat frag- the role of dominance. Evolution 47:949– mentation for plants. TREE 11:413–418. 957. Young AG, Brown ADH, Zich FA. 1999. Genetic Wright S. 1931. Evolution in Mendelian popu- Structure of Fragmented Populations of the lations. Genetics 28:114–138. Endangered Daisy Rutidosis leptorrhynch- Yates CJ, Coates DJ, Elliott C, Byrne M. 2007. oides. Conservation Biology 13:256–265.

19 Investigating genetic factors behind the decline of a threatened plant species

Tack!

Björn Widén. Jag har framför allt uppskattat ditt intresse för de spår som historien lämnar i landskapet, och som står att ¤nna ända in i en till synes obetydlig blommas DNA. Vilket underbart doktorandprojekt för den som en gång var på väg att bli arkeo- log. Utan dig hade jag fortfarande inte vetat vad en fältnocka är. Och tack för hjälp med allt från omplantering till manusskrivande.

Stefan Andersson. Dina råd är alltid välkomna och realistiska, och du kan se det intressanta i resultat som en nybörjare ¤nner nedslående. Du är en pärla!

Mikael Hedrén, för goda råd och för din förmåga att göra trevligt småprat till en skön konst.

David Ståhlberg, för att du var en inspirerande arbetsrumskompis, för all fågelsång du har lärt mig, och för alla mysiga politiska diskussioner, samt inte minst för att du ställde upp som min fältassistent på Jylland! Och tack för all hjälp med manusläs- ning i doktorandlivets slutskede.

Linda-Maria Mårtensson, för ut¥ykter, ¤kor, middagar, ¤lmkvällar, avspänt kattum- gänge, samtal om allting och ett starkt moraliskt stöd.

Ullrika Sahlin, för din humor och dina infall. För statistikhjälp, för uppmuntran och för din förmåga att kombinera fantasi och fakta.

Jakob Sandberg, för att du är så mån om viktiga detaljer, för alla vänliga ord och för din hjälpsamhet.

So¤e Nordström. Det var ett nöje att samarbeta med dig på sluttampen. Du är en kvinna som vet var man sätter punkt, och du glömmer aldrig att berätta när du tycker att nånting är bra.

Louise Hathaway för hjälp med det engelska språket, och för alla gratisprogram du har tipsat mig om, samt för glada YouTube-klipp och tips om ¤lmer och böcker som jag inte visste fanns.

Jane Jönsson, för att du dök upp den där hösten när allt skulle planteras om och ingen tid fanns!

Eman Soubani. Du har också planterat om en försvarlig mängd fältnockor – det kan inte underskattas!

Frida Rosengren. Du mätte alla frö och fröpenslar till sista manuset!

Maarten Ellmer. Så bra att du fanns i rummet snett emot under de sista veckorna. Annars hade det varit bra ensamt här!

Åsa Lankinen. Du har lyssnat när det har varit som kämpigast utan att komma med pekpinnar!

20 Investigating genetic factors behind the decline of a threatened plant species

Triin Reitalu. Tack för alla tips inför Estlands-resan. Utan dig hade jag varit helt bortkommen i det lilla landet där fältnockorna är så stora!

Pernilla Göransson och Frida Andreasson. Synd att ni försvann när jag var ungefär halvvägs. Det var så trevligt när ni var här.

Johanna Eneström. Du har lovat att lära mig stå på händer. Kvarstår det erbjudan- det?

Io Skogsmyr, för ditt demokratiarbete och dina ansträngningar att få alla att känna sig välkomna.

Galina Garkavia. Du har alltid praktiska lösningar på besvärliga problem. Du är det nav kring vilket hela avdelningen roterar.

Maj-Lis Gernersson. Du glömmer aldrig att fråga om det har ordnat sig när du vet att man har bekymmer.

Staffan Bensch, för AFLP-relaterad akuthjälp.

Bengt Jacobsson. För din hjälp i växthuset och din välvilja.

Maie Jeeser, Toomas Kukk, Triin Reitalu, Kaja Riiberg, and Elle Valtna who helped me to ¤nd Estonian lood-ristirohi-populations. And particularly Andres Miller and Elle Roosaluste for helping me reach the Hiiumaa islets.

Alan Knapp, Tony Mundel, David Pearman, and Sharon Pilkington who helped me localise Field Fleawort-populations in England.

Hans Henrik Bruun. Tack för att du hjälpte mig hitta de danska Bakkefnokurt-popu- lationerna.

Mamma, för att du var fältassistent och gott sällskap åt mig både i Estland och i England, för att du är smart och rolig, och har bidragit med så goda gener – kanske det viktigaste bidraget till den här avhandlingen.

Pappa. Tack så väldigt för layouthjälpen!! Och för språkgranskning. Du kan också vara smart och rolig när du sätter den sidan till. Och gener har jag ju fått av dig med, tack så mycket!

Min bror Mårten och min svägerska Anneli, det är alltid skönt att få prata med er, och mina brorsöner Edvin, Anton, Eelis och Henri för att ni är roliga, söta och rara!

Och tack till alla andra – vänner, släktingar, kollegor och träningskompisar!

I Inbreeding depression cannot explain the rapid decline

24 Inbreeding depression cannot explain the rapid decline

Inbreeding depression cannot explain the rapid decline of the self-incompatible perennial Tephroseris integrifolia in Sweden Kerstin Isaksson and Björn Widén, Department of Ecology, Section Plant Ecology and Systematics, Lund University, Sweden.

Abstract Habitat loss, change, and fragmentation are serious threats to many species. Historical docu- ments indicate that the rare ¤eld ¥eawort, Tephroseris integrifolia (L.) Holub (Asteraceae), in Skåne, southern Sweden, has decreased from about forty populations in the mid 19th century to six populations in the early 21st century. This decrease in population number can to a large extent be explained by the conversion of its natural habitat, nutrient-poor calcareous grassland, to arable ¤elds and forest plantations. A census of extant populations from 1980 to 2009 revealed a dramatic decrease in population sizes during the early 1990s, despite far-reaching conservation measures, and even extinction of two populations in the late 1990s. Here we investigate if this decline can be explained by inbreeding depression as a consequence of reproductive isolation. However, we found only weak signs of inbreeding depression, which consequently cannot explain the drastic decline of the species. We discuss possible alternative causes of the decline. Seedling deaths and a low seedling-recruitment seem to be strongly connected with drought, grazing, and mollusc herbivory, and are probably connected to a restricted seed-dispersal. We thus conclude that the most probable causes of the decline are a combination of lack of suitable habitat and of stochastic external threats.

Keywords: Population census, inbreeding depression, population structure, threatened species, extinction

Introduction Davies 1976, Ouborg 1993, Willi et al. The loss of suitable habitats as a conse- 2006). Habitat fragmentation might also quence of human activities has led to de- lead to an increased exposure to diseases creasing population sizes of many for- (Groppe et al. 2001, Mbora & McPeek merly common species, as well as local 2009), to the invasion of new species extinctions of whole populations (Fischer (Cillers, Williams, & Barnard 2008), and & Stöcklin 1997, Luijten et al. 2000, to altered water and soil conditions Matthies et al. 2004). Habitat fragmen- (Camargo & Kapos 1995, Cramer & tation has also led to increasing isolation Hobbs 2002) as a consequence of a larger of populations and individuals, which in edge-to-surface ratio. Habitat change, the case of sessile organisms, such as isolation and edge-effects together affect plants, often causes complete or partial species composition (Kiviniemi & Eriks- reproductive isolation (Snaydon & son 2002) as well as population de-

25 Inbreeding depression cannot explain the rapid decline

mography and reproductive success (e.g. lings may also indicate that a large num- Widén 1993), and might in the end cause ber of homozygotes do not survive until genetic depletion (e.g. Van Rossum et al. the adult stage, or even that no or ex- 2002). tremely few seedlings survive, regard- The spatial genetic structure of natu- less of their genetic constitution, if the ral populations is in¥uenced by proces- environment is so altered that recruit- ses such as genetic drift (Shah et al. ment is made impossible. The remaining 2008), gene ¥ow (Young, Brown, & Zich adult population may then consist of old 1999, Gaudeul & Till-Bottraud 2008), individuals, which means it is a relict of and natural selection (Snaydon & Davies a formerly large and thriving population 1976). In plant populations, low genetic (Luijten et al. 2000, Van Geert, Van variation within fragments and high Rossum, & Triest 2008). Excess variation variation between fragments have been in alleles under balancing selection such observed, and are believed to be the as self-incompatibility-alleles or, to a cer- results of genetic drift and restricted tain extent, major-histocompatibility- gene-¥ow (Shah et al. 2008, Liu et al. complex-alleles, over putatively neutral 2009). However, in many studies, an un- genes are also an indication of loss of expectedly high genetic variation has variation through genetic drift, but at been found in supposedly isolated po- the same time examples of mechanisms pulation fragments or small populations, counteracting the detrimental effect of and often also an unexpectedly low gen- drift (Luijten et al. 2000, Mable & Adam etic variation between fragments or rela- 2007, Vandepitte et al. 2007). If small tively isolated populations (Young, populations consist mainly of adult Boyle, & Brown 1996, Young, Brown, & ¥owering plants, they might be survivors Zich 1999, Marquardt et al. 2007, from a much larger population and hence Williams et al. 2007, Yates et al. 2007, contain an unexpectedly large genetic Yoshioka et al. 2007, Gaudeul & Till- variation (e.g. Luijten et al. 2000). Bottraud 2008). One explanation is a In studies of the effects of habitat high gene-¥ow between fragments or fragmentation, it is important to cor- populations, particularly in wind-pollin- rectly estimate population sizes. Popula- ated species (Williams et al. 2007), or tion size seems to be crucial for survival gene-¥ow from undetected individuals of of rare and threatened species (Matthies the same species (Lawrence et al. 2008). et al. 2004, Münzbergová 2006). To Under some circumstances, fragmenta- estimate past plant population sizes is tion even seems to increase gene-¥ow dif¤cult. Historical development of indi- (Young, Brown, & Zich 1999). However, if vidual populations is inherently dif¤cult the time since fragmentation is short to assess, but the general trends in popu- compared to the generation-time of the lation development can be inferred from studied species, it might not yet be pos- past and present distribution maps (e.g. sible to observe the genetic consequences Luijten et al. 2000). Site area (Ouborg of fragmentation (Young, Boyle & Brown 1993) or mean cover (Fischer & Stöcklin 1996, Williams et al. 2007, De Almeida 1997) are often substitutes for popula- Viera & de Calvalno 2008). This is fur- tion size. Direct counts of numbers of in- ther con¤rmed by studies where the off- dividuals in plant populations are al- spring have a lower heterozygosity than most lacking before the 1980s (Matthies their parents (Lawrence et al. 2008, Van et al. 2004). Geert, Van Rossum, & Triest 2008), indi- In the present study we investigate cating that fragmentation will take its possible inbreeding depression in the toll on genetic diversity in the near fu- rare plant Tephroseris integrifolia – a ture. However, a large discrepancy in species which has continually declined in genetic variation or heterozygosity be- Sweden during the past 140 years as a tween established individuals and seed- consequence of habitat loss and fragmen-

26 Inbreeding depression cannot explain the rapid decline

tation, as well as of decreased grazing, 1993). No soil seed bank is accumulated but which is still declining, despite (Widén 1987). In an experimental study considerable conservation measures. We Widén (1993) found a weak trend of in- use a combination of historical records breeding depression on early stages of (herbarium specimens and literature the life cycle (germination percentage data) and empirical data from the last 30 and germination speed), but no correla- years to show the decline in population tion between population size and levels number and size of this rare and endan- of inbreeding depression. The plant gered plant species in Sweden. We per- grows in calcareous, grazed semi-natural formed crossing experiments to explore if grasslands. It is sensitive to competition inbreeding depression could be a key fac- from other plants, thus grazing increases tor explaining the rapid decline in popu- seed germination and seedling survival – lation size and number. The questions however, grazing early in the season we want to answer are: tends to reduce reproduction, since both 1. Is inbreeding depression a key fac- ¥owers and seeds are grazed (Widén tor in the decline of T. integrifolia in 1987). Sweden? T. integrifolia has a disjunct distribu- 2. Could inbreeding depression be a tion with the westernmost populations threat to T. integrifolia in the fu- occurring in Britain and the easternmost ture? in Japan (Smith 1979).

Tephroseris integrifolia in Sweden Material and methods In Sweden T. integrifolia is recorded only Study species in the southernmost province, Skåne, Tephroseris integrifolia (L.) Holub and the ¤rst record is from 1623 (Anders- (Asteraceae) is a perennial plant, 10–50 son 1944). The decline of the species be- cm high, with 3–10 leaves in the basal gan already in the late 19th century rosette, and with 1–3 leaves on the (Nilsson & Gustafsson 1977). The basic ¥owering stem. Usually one major and a threat to the plant has been habitat frag- few minor rosettes develop on an almost mentation caused by changes in land- vertical rhizome. The shoot producing use, but the present-day decline in popu- the ¥owering stem dies after seed dis- lation size may also be connected with persal, and one of the minor rosettes deterioration of habitat quality. Many becomes dominant, or adventitious buds habitats of the species have been ex- on the rhizome develop into new ro- posed to arti¤cial fertilization, ceased settes. The ¥ower heads (1–4, sometimes grazing, or impact from invasive species more) are arranged in a corymb. The (Widén 1987 and unpublished). plant ¥owers in May to June, and the In the present study we use plant number of ¥owering plants per popula- material from three populations repre- tion varies considerably between years senting the two geographical areas with (Widén 1993). Plants rarely ¥ower in two extant populations (NE and SE Skåne, successive years, and ¥owering is par- respectively). ticularly poor the year after a heavy drought. The primary pollinators are Benestad (SE Skåne) Diptera (mostly syrphid ¥ies). The plant The Benestad population is situated on a is self-incompatible and has wind-dis- south-west facing slope within the Bene- persed fruits (achenes, hereafter called stads backar nature reserve with grazed, seeds) with pappi (Widén 1987). Seed set unfertilised pastures on calcareous soils. is density dependent, increases with Several hundred years of limestone population size, and shows an optimum quarrying has created the present-day for individuals ¥owering during the peak topography, which consists of a mosaic of of the ¥owering period (Widén 1991, dry hills with species-rich grassland ve-

27 Inbreeding depression cannot explain the rapid decline

getation separated by wet-land vegeta- 1976, 2003). T. integrifolia grows on the tion fed by ground-water on plateaus and top and the slopes of the hills in rather in valleys between the hills (Regnéll well-de¤ned patches of a few to 50 m2,

(a) (b)     

Fig. 1. Schematic map of the spatial distribution of patches occupied by Tephroseris integrifolia at Benestad (a) and Grödby-Råby (b). Patches with plants recorded for the last time during the 1980s have been marked with open symbols. Gray indicates extinction during the 1990s and black indicates records of flowering plants 2000–2009. Figures indicate the position of seed samples and letters denote patches mentioned in Table 1. separated from each other at a distance tegrifolia (Grödby N). of 10 to 50 m (Fig.1a). Råby: The Råby population is situated 300 m further to the north-east of the Grödby–Råby (NE Skåne) Grödby site. The vegetation is trivial and Grödby: Until the late 1950s, when a the area has obvious remains from old large part of the Grödby area was plan- arable ¤elds (piles of stone). According to ted with pine, the Grödby population the local inhabitants, potato was culti- was one of the largest known T. integri- vated here during the 1950s. folia populations in Sweden (Nilsson & Grödby N is closer to the Råby popu- Gustafsson 1977). The plantation divi- lation than to Grödby S (Fig.1b). It could ded the Grödby population into two be argued that it is more logical to regard parts, referred to as Grödby S and Grödby S, Grödby N, and Råby as three Grödby N, about 400 metres apart. In sub-populations of the same population, the early 1980s plants were scattered than to consider Grödby and Råby as two around the pine plantation (Fig.1b). The separate populations. However, due to area south of the pine plantation (Gröd- the relatively recent connection between by S) is a species-rich grassland with sev- Grödby S and Grödby N, we have chosen eral rare species indicating a long con- to regard Grödby as one population tinuous grazing management. To the which has recently suffered severe frag- east the pine plantation borders on mentation, and the Råby population as a arable ¤elds, and to the west and north separate population – since we are not there are grazed pastures with a trivial interested in the genetic similarities or vegetation (Widén, personal observa- differences between populations, but tion), with scattered patches with T. in- rather the relationship between physical

28 Inbreeding depression cannot explain the rapid decline

Table 1. Data from samples in natural populations 2000. Mean (SD) and sample size, n. Patch number refers to Figure 1. Number of stems refers to the number of flowering stems within the sampled area (patch).

Population Patch No of Plant Number No of Fruit Achene Germination stems height of heads florets set weight rate /head 10 -3 g

Benestad 1 33 14.1 2.9 52.9 0.29 0.54 0.63 (2.97) (0.74) (25.66) (0.295) (0.147) (0.443) 15 15 15 15 12 12 a 1 21 3 65 0.02 0.62 1.0 (-) (-) (-) (-) (-) (-) 1 1 1 1 1 1 2 18 23.2 3.9 75.9 0.47 0.62 0.77 (5.91) (1.73) (19.26) (0.209) (0.116) (0.328) 14 14 14 14 14 14 b 5 20.0 3.0 72.0 0.07 0.67 0.67 (1.41) (0) (21.21) (0.098) (-) (-) 2 2 2 2 1 1 3 20 16.5 2.6 62.2 0.17 0.53 0.72 (4.37) (1.17) (21.6) (0.159) (0.126) (0.354) 11 12 12 12 11 11 c 3 18 5 65 0 - - (-) (-) (-) (-) 1 1 1 1 Grödby S 1 132 20.3 4.2 76.9 0.64 0.47 0.88 (5.42) (1.67) (13.59) (0.231) (0.072) (0.244) 27 27 27 27 26 26 2 43 24.2 5.3 87.2 0.47 0.47 0.96 (4.92) (1.75) (16.62) (0.227) (0.054) (0.259) 13 13 13 13 13 13 3 82 20.5 4.6 82.1 0.42 0.48 0.87 (6.68) (1.81) (20.06) (0.223) (0.103) (0.181) 22 22 22 22 21 21 Grödby N 4 16 15.2 5.2 85.1 0.13 0.42 0.86 (2.60) (1.66) (21.51) (0.117) (0.071) (0.204) 11 11 11 11 10 10 5 24 20.8 4.8 79.5 0.15 0.42 0.80 (2.75) (2.48) (12.52) (0.130) (0.071) (0.304) 11 11 11 11 9 9

Råby 1 153 19.2 4.6 85.2 0.49 0.47 0.90 (4.64) (1.78) (17.04) (0.256) (0.065) (0.109) 14 14 14 14 14 14 2 26 20.8 6.1 75.0 0.34 0.55 0.80 (5.02) (3.02) (15.24) (0.212) (0.079) (0.326) 11 11 11 11 10 10 3 20 21.0 5.4 86.7 0.27 0.57 0.82 (6.29) (1.71) (14.11) (0.287) (0.128) (0.319) 10 10 10 10 9 9 population structure and genetic varia- in the past, we recorded date and locality tion. of ¤ndings accounted for in published records collected in the ¥oristic inventory Investigation of historical popula- of Skåne (Tyler et al. 2007) as well as of tions herbarium specimens in the Botanical To estimate the number of T. integrifolia Museum at Lund University. Since the populations in Sweden at different times exact positions of earlier ¤ndings are

29 Inbreeding depression cannot explain the rapid decline

often vague, whereas parish is always experimental plants in this study, given, we record presence and absence of samples were rarely collected more than T. integrifolia in parishes for this investi- 10 metres apart within the same patch. gation. We make the following two as- The positions of patches occupied by T. sumptions: 1. As T. integrifolia is a rela- integrifolia have been indicated on maps tively slow colonizer; we assume that (Fig.1). GPS was used to determine the ¤ndings of T. integrifolia in a certain par- centre of small patches and the centre ish during this period (1800–2009) indi- and outline of large patches. Because of cate presence of the species in that par- the irregular ¥owering of the species it is ish since 1800. 2. When ¤ndings are no dif¤cult to determine year when the longer made in a parish, we assume T. plant has gone extinct in a patch. Here integrifolia to have gone extinct in that we report the decade when the last parish. We believe that this gives a ¥owering individual was observed in a rather good estimate of when the species patch. has disappeared from different parts of The sampling design was based on the Skåne. We also measured the height of structure of the Benestad population, all herbarium specimens and counted which has a clear, patchy structure. the ¥ower heads, in order to ¤nd signs of From the other two populations we inbreeding depression prior to extinc- sampled from areas (»arti¤cial patches») tion. Height and number of ¥ower heads of the same size as the Benestad patches, were analysed separately with ANCOVA, demarcated within more continuous testing for the effects of administrative areas or larger patches, and approxi- county district (Swedish härad) and col- mately on the same distances from each lection-year. other as the Benestad patches. From Be- nestad we collected seeds in three well- 30 year census de¤ned and well-separated patches (1, 2, From 1980 onward one of us (BW) esti- and 3). The positions and areas of these mated population sizes by counting the patches have been rather stable since annual number of ¥owering stems in na- 1980. In Grödby S, three arti¤cial tural populations. The counting was al- patches (or more strictly, sampling ways performed early in the ¥owering areas), 1, 2, and 3, were demarcated and season, before any grazing had reduced sampled from. Grödby N has a patchy the number of ¥owering stems. If grazing structure, although not as clear as at Be- had occurred, attempts were made to in- nestad – the structure varies between clude grazed stems. years and scattered plants appear between patches. Seeds were sampled from Sampling of experimental plants two patches (4 and 5). From the Råby Populations of T. integrifolia often have a population, which mostly consists of patchy structure, with a varying dis- three large patches (with scattered, ir- tance between the patches. In this study regularly ¥owering individuals between we use three populations with contrast- the patches), we demarcated and collecting population structure and patch dy- ed seeds from one arti¤cial patch within namics (Fig.1). As is clear from Table 1, each of the larger patches. plants growing alone, or very few to- Seeds were collected from one ¥ower gether (patches a, b, and c at Benestad) head per plant during the fruiting period failed to set seed. These plants grew in mid June. We determined seed set as more than 10 metres away from other the proportion of developed achenes of plants, indicating that the maximum the total number of ¥orets per ¥ower pollination distance is less than 10 head. The positions of the sampled metres. We therefore de¤ned a patch as a plants were determined to the nearest group of plants which grow more than 10 dm in a coordinate system for each popu- metres away from other plants. For the lation. Plant size, and seed weight (aver-

30 Inbreeding depression cannot explain the rapid decline

age weight of 20 seeds per mother) were had the same ¤eld parent). also recorded (Table 1). The main sam- 2. Within patch crosses (i.e. ¤eld par- pling was done in 2000, and an addition- ents were not identical, but grew al sampling was done at Råby and at within the same patch). Grödby N in 2001. Since plants rarely 3. Between patch crosses (i.e. ma- ¥ower during two successive years, we ternal parents grew in the same assume that we have not sampled from population but in different patches). the same plant twice. Seeds were sown In some analyses we separated be- indoors in the autumn the same year as tween crosses 3a. between close sampled, and the progeny plants were patches (less than 150 metres apart) kept outdoors during the winter. and 3b. between distant patches (400-500 metres apart) – i.e. crosses Plant size and seed set in natural between Grödby S and Grödby N. populations 4. Between populations crosses. Cros- We compared plant-height and seed-set ses between plants from different in the ¤eld in the three populations with populations (Benestad and Grödby). ANOVA (with population as a ¤xed factor These offspring were only analysed and number of ¥owering plants per with respect to germination, sur- patch as a covariate). In the same way vival, and ¥owering. we analysed weight and germination rate of the collected seeds. Since seed-set The plants to be pollinated were kept in and germination rate are both ratios, we an insect-free greenhouse environment. transformed them with the logit-func- We performed the cross-pollination by tion (see below under Germination). We gently rubbing two ¥ower heads against also analysed the relationship between each other on two different occasions number of ¥owering plants and the por- with a one- to three-day interval. The tion of plants which had received com- ¤rst cross-pollinations were carried out patible pollen – counting seed-set below when ray ¥orets and the outer tubular 0.02 as 0, and seed-set above 0.02 as 1 (cf ¥orets were receptive, and the second Isaksson paper II). This analysis was when most ¥orets in the two ¥ower performed with a generalised linear heads were fully developed. model, with seed-set given a binary-pro- bit distribution. Cultivation of offspring The rapidly declining viability of seeds Crossing experiments makes it necessary to sow seeds within Most of the cross-pollinations were per- 1.5 years after harvest. The seeds from formed in 2002 between plants from the the 2002 and 2003 cross-pollinations 2000 seed collection. Additional pollina- were sown in October 2003. Plants de- tions were performed between the Råby rived from this sowing are referred to as plants in 2003, both from the 2000 and cohort 2003 below. Only plants of cohort 2001 seed collections. However, we never 2003 were used for all experiments. The cross-pollinated two Råby plants from seeds from the 2004 cross-pollinations different collecting years. Additional were sown in July 2004 (cohort 2004). We cross-pollinations were done on the Be- sowed 20, or all seeds, if they were less nestad and Grödby S and N plants in than 20, from each harvested ¥ower 2004. head on Petri dishes. After germination they were transferred to 5×5 cm2 pots Crossing categories and distances and planted in standard soil. The cross-pollinations of the plants were In all c.8175 seeds from 503 crosses divided into different categories based on were sown in 2003. Of the surviving off- the positions of the ¤eld parents: spring, plants from about 300 crosses 1. Between sibs crosses (i.e. the plants were saved for morphological analyses.

31 Inbreeding depression cannot explain the rapid decline

Five seedlings from each family were value 0. All cases were pooled within transplanted to 10×10 cm plastic pots population. This was analysed for (one seedling per pot) in the greenhouse. the populations separately with χ2 After about a month, the pots were with respect to crossing-category. transferred to a common garden out- 2. Germination rate – that is, the doors where they were kept during the number of germinated seeds divided winters. by the number of sown seeds from In July 2004 about 5500 seeds from one ¥ower head. Since germination 320 crosses were sown. Of the surviving rate is a ratio, it was transformed offspring, plants from about 100 crosses with the logit function: were saved for morphological analysis. Logit(germination rate) The plants in cohort 2003 had by =log((germination rate/(1-ger- September 2004 developed new rosettes, mination rate)) and from each genet, one or two rosettes Germination rate=0 was excluded of about the same size as the 2004 from this analysis. Germination seedlings were transplanted to separate rate was analysed with ANOVA, 10×10 cm2 pots. At the same time the where ¤eld paternal grandmothers seedlings in cohort 2004 were also trans- were nested within population, and planted to 10×10 cm2 pots (¤ve from each crossing-category, population, and family). All the plants were kept in the interaction between crossing cat- green house for two months. Where two egory and population were ana- rosettes from the same genet were lysed. available, one was selected for ¤eld- 3. Germination time – that is, average transplantation, and transplanted to the germination date of the germinated site of an extinct population (Lövhall, see seeds. This was analysed with Fig.5) in November 2004. All other ANOVA, where ¤eld paternal grand- plants were at the same time transferred mothers were nested within popu- to a common garden. lation, and crossing-category, population, and interaction between Germination crossing-category and population Germination in 2003 was monitored were analysed. from October 14, when the seeds were sown, till November 19. Germination in Flowering and survival 2004 was monitored from July 5, when Survival was recorded on two occasions the seeds were sown, till August 2. Some for plants in the common garden experi- seeds were sown in the same year when ment, in April 2005, and in November they were harvested and some in the 2005 (i.e. before and after ¥owering). following year. Since germination de- Flowering was recorded in summer creases drastically with time, we made a 2005. For the ¤eld experiment (plants strict separation between seeds sown transplanted to the site of the extinct soon after harvesting, and those stored population Lövhall), survival, ¥owering, from the previous year (hereafter called and number of ¥owering stems were re- young and old seeds, respectively). corded in spring 2005. Both ¥owering We analysed the germination data in and survival were analysed with χ2-tests. three different ways: For the common garden experiment, we 1. Germination or not – that is, analysed survival of winter and survival whether any seeds from one ¥ower of summer, separately and together, as head germinated. Flower heads well as survival after ¥owering. with less than 5 ¤lled seeds were excluded from the analysis. At least Morphology one germinated seed gave the value The following morphological variables 1, and no germinated seeds gave the were scored on plants in different cros-

32 Inbreeding depression cannot explain the rapid decline

sing-categories: height of ¥owering stems, number of stem-leaves, number of ¥ower heads, number of ray-¥orets, length of ray-¥orets (average based on three ¥orets), width of ray-¥orets (average based on three ¥orets), and width of ¥ower head disc. Flower-characters were scored on ¤rst ¥owering ¥ower head. All morphological characters were analysed separately with one-way ANOVAs, where ¤eld-grandmother was nested within population. We checked the effect of crossing-category, population, ¤eld grandmother, the interaction of crossing category and population, and the interaction of crossing category and ¤eld Fig. 2. Map of the province Skåne with par- grandmother. ishes that have been reported to harbour populations of Tephroseris integrifolia since Statistics the middle of the 19th century. Light grey indicates extinction before 1900, dark grey All statistical analyses were performed indicates extinction during the 20th century with SPSS 16.0 (SPSS Inc., Chicago, IL). and black indicates extant populations after 2000. (Two dubious findings have been excluded.)

Results Historical populations A maximum number of 42 local populations in Skåne can be identi¤ed from herbarium material and literature data. The number of parishes with records of T. integrifolia decreased from 26 in the early 19th century to 4 in the early 21st century (Fig.2 and 3). The decline in number of records was most pronounced in the late 19th century (Fig.3). The estimated number of local populations was 15 at the turn of the 20th century, and six at the turn of the 21st century. In the 19th century, the species was recorded in both the western and the eastern parts of Skå- Fig. 3. Number of parishes in Skåne with ne, but it has now disappeared from the known populations of Tephroseris integri- west, and from most of the original east- folia from 1835 to 2009. ern localities. 30 year census According to the measures of the her- When the census of populations started barium specimens, sizes of T. integrifolia in 1980, seven populations were known, plants have ¥uctuated a lot. However, most of which were included in the an- there is a general trend that the plants nual census. The seventh population, are taller the later they were collected Edenryd, was checked most years, but a (p<0.05, Fig.4), while at the same time regular count of number of ¥owering the number of ¥ower heads does not stems did not start until 1999. At one change with time (p>0.05, data not site, Grödby, the spatial distribution of T. shown). integrifolia was unclear in 1980, and new

33 Inbreeding depression cannot explain the rapid decline

populations Åby, Edenryd, and Kverre- stad had, like the present-day Grödby population, a sub-population structure with aggregations of patches separated by distances of >100 metres. The number of patches at each site has decreased during 1980–2009. In the 1980s, many patches had high densities of T. integrifolia plants. In Tosteberga, Edenryd, and Grödby, patches that still had surviving plants in 2009 tended to be associated with large stones and shrubs. The distribution of plants in the core area of the eastern part at Grödby S had been more or less continuous since the early 1980s, Fig. 4. Plant size of herbarium specimens in though the irregular ¥owering of the Lund Botanical Museum sampled since the early 19th century in different districts (hä- plants has created a patchy annual rad). Plant height (cm), differences be- structure of ¥owering individuals tween years and districts are significant at (Fig.1b). The population structure in p≈0.01. Interaction between year and dis- Grödby N is patchy, but patches are trict is significant at p=0.012. rarely separated by more than 10 metres patches of the plant were detected dur- in the core area of the eastern part. At ing the following years. These are pre- Råby, plants tend to be clumped together sented as Grödby N (see Fig.1b) in Fig.5. around stones, but scattered individuals The eighth population, Råby, was redis- have often occurred in the gaps between covered in 1983 but a regular counting of patches. ¥owering stems did not start until 1999. At Benestad about 20 patches have The number of ¥owering stems varied been identi¤ed since 1980 (Fig.1a). much between populations, with maxi- Flowering in some of the patches has oc- ma ranging from 50 at Åby to 8000 at curred with several years interval be- Tosteberga (Fig.5). The numbers ¥uctu- tween 1980 and 2009. The majority of ated dramatically between years in all the patches are very small and none of populations. In the largest population in them cover any large area, and due to the the 1980s, Tosteberga, a general trend topography at the site, they are all well could be found that can also be traced circumscribed. with some modi¤cations in the other populations. From 1980, the numbers in- Plant size and seed set in natural creased and reached a maximum in populations 1982, and then decreased to a minimum In the ¤eld, seed-set was signi¤cantly in 1984, after which a new increase lower in the Benestad population than in started with a maximum in 1988, and the Grödby population (p=0.023). How- again a minimum in 1991. After 1990, ever, the effect of population was no the number of ¥owering plants ¥uctu- longer signi¤cant (p=0.8) when number ated between much lower levels than ob- of ¥owering plants per patch were taken served during the 1980s in all popula- into the analysis – which was signi¤cant tions. Two populations, Lövhall and Åby, at p=0.0000002. The plants in the three have gone extinct with the last record of small Benestad patches a, b, and c ¥owering stems in 1997 (Fig.5). (Fig.1a) had almost zero seed-set. The plants in Benestad patch number 2 had Spatial structure an unusually high seed-set (on average All populations have a patchy spatial 0.47) considering the relatively low structure, and in the early 1980s the number of ¥owering plants (18), com-

34 Inbreeding depression cannot explain the rapid decline

Fig. 5. The annual number of flowering stems of Tephroseris integrifolia at sites with extant populations in 1980 or later. Flowering was observed at Edenryd, Grödby N, and Råby since the 1980s, but a regular counting of flowering stems did not start until 1999.

35 Inbreeding depression cannot explain the rapid decline

ses rendered seeds with higher germination rate than the rest, and for the Grödby population, where seeds from between-patch crosses had the lowest germination rate. For the Råby-population, the trend was similar (p=0.043) to that in the other two populations (Fig.7a).

Germination rate Germination ability of T. integrifolia seeds decreased rapidly with time; 78% of young seeds germinated, while 30% of old seeds germinated. Germination rate was thus relatively low for old seeds, and Fig. 6. Relation between seed set and num- we could ¤nd no consistent difference be- ber of flowering stems of Tephroseris integri- tween crossing categories (Benestad: folia at each patch in 2000. Open symbols denote Benestad and filled symbols Gröd- p=0.60; Grödby: p=0.76; Råby: p=0.14 – by-Råby. data not shown). However, for young seeds there was a signi¤cant difference pared with other similar-sized patches in in germination-rate between crossing- this study (Table1, Fig.6). Plants at categories for the Benestad plants – be- Benestad were on average shorter than tween-patch crosses gave a signi¤cantly the others, but the differences were not higher germination rate than sib- and signi¤cant (p=0.083 compared with within-patch crosses, and the Grödby Grödby and p=0.22 compared with Rå- plants – sib-crosses gave a signi¤cantly by). There was no signi¤cant difference lower germination rate than the rest, but in seedweight (p=0.26) or germination not for the Råby plants, though they rate (p=0.095) between populations, and seemed to show the same trend as plants they were not related to number of from the other two populations (Fig.7b). ¥owering individuals per patch (p=0.125 and 0.246, respectively). There Germination time was a signi¤cant relationship between For old seeds we found no signi¤cant re- number of ¥owering plants per patch lationship between germination time and portion of plants which had received and crossing category (Benestad: p=0.80; compatible pollen (p=0.008). Grödby: p=0.57; Råby: p=0.46 – data not shown). However, there was a signi¤cant Germination difference in germination time for young Germination or not seeds derived from Benestad and Gröd- When we only investigated whether any by, with the sib-cross-seeds germinating seeds from one head had germinated, later than seeds from other cross-pollin- 93% of young seeds germinated (97% if ations, but not for seeds from Råby, only cases with at least 5 sown seeds which showed a similar trend, but no sig- were included) while 68% of old seeds ni¤cance (Fig.7c). germinated (75% if only cases with at least 5 sown seeds were included). Ger- Flowering mination rate was high for young seeds Plants in cohort 2003 were too small and no difference between categories when transferred to the experimental could be established. For old seeds, there beds outdoors to be able to ¥ower in was a signi¤cant difference between 2004, while plants in cohort 2004 had crossing-categories for the Benestad grown large enough during their ¤rst population, where between-patch cros- year to ¥ower in 2005. The overall fre-

36 Inbreeding depression cannot explain the rapid decline

(a) (c)

(b)

Fig. 7. Germination of Tephroseris integrifolia seeds for different crossing categories (1=sib cross, 2=within patch cross, 3=between patches cross, 3b=between distant patches cross). Figures above bars indicate cample sizes. (a) Germination or not for old seeds. Differences between crossing categories are significant at p=0.002 for Benestad, at p<0.001 at Grödby and marginally significant (p=0.043) at Råby. (b) Germination rate for young seeds. Differences between crossing categories are significant at p=0.002, and differences between populations at p<0.001. (c) Germination time of young seeds. Differences between crossing categories are significant at p=0.001 (for Benestad p=0.007, for Grödby p=0.006, for Råby p=0.33), and differences between populations at p<0.001.

37 Inbreeding depression cannot explain the rapid decline

quency of plants in cohort 2003 that (a) ¥owered in 2005 was about 40% with signi¤cant differences between crossing categories, sib-crosses showing the lowest tendency to ¥ower (Fig.8a). The differences in ¥owering frequency between crossing categories were signi¤- cant at Grödby and Råby, but not at Be- nestad. At Råby, within-patch crosses gave higher ¥owering-frequencies than between-patch crosses (Fig.8a). The tendency to ¥ower in 2005 was slightly lower in cohort 2004 (Fig.8b) than in cohort 2003 (Fig.8a). There was a signi¤cant difference in ¥owering be- (b) tween crossing-categories for the Bene- stad plants, with the highest ¥owering frequency for the between-patch crosses, and between-sibs and within-patch crosses rendering plants with almost equal ¥owering frequencies, but not for the Grödby plants, which had very low ¥owering frequencies for all crossing categories. Flowering frequency of plants transplanted to the ¤eld differed signi¤- cantly (p<0.05) between crossing categories (Fig.8c). (c) Survival For the plants in cohort 2003, we could detect no signi¤cant differences in survival during the winter 2004/2005 between categories, probably because survival generally was so high (data not shown). For the survival of the entire year 2005, there was a signi¤cant difference between crossing categories for the Grödby plants, where plants from between-patch crosses had a higher survival rate than plants from the other crossing categories and Råby, where plants from withinpatch crosses had a higher survival rate than plants from Fig. 8. Flowering frequency of Tephroseris in- sib-crosses and between-patch crosses, tegrifolia for different crossing categories (1=sib cross, 2=within patch cross, 3=be- but not for Benestad (Fig.9a). tween patches cross, 4=between popula- For the plants in cohort 2004, which tions cross). Cf. fig. 7 experienced their ¤rst winter in 2004/ (a) Flowering frequency in cohort 2003 in 2005, there was a signi¤cant relation- cultivation 2005. ship in survival between crossing-cat- (b) Flowering frequency in cohort 2004 in cultivation 2005. egories among the Benestad plants, (c) Flowering frequency in the field 2005 where between-sibs crosses gave the of cohort 2003 transplanted to the field in lowest and between-patch crosses the November 2004.

38 Inbreeding depression cannot explain the rapid decline

(a)

(b) Fig. 10. Height of flowering stems of Teph- roseris integrifolia in cultivation 2005 (both cohort 2003 and cohort 2004) in different crossing categories (cf. fig. 7 & 8).

highest survival, but not among the Grödby plants (Fig.9b). For the plants that ¥owered in 2005, there was no signi¤cant difference in survival between categories, probably because the sample was too small (data not shown). Survival of ¤eld transplants was the lowest for between-sibs crosses, and in- (c) creased with distance between ¤eld- grandmothers (Fig.9c) – though Grödby and Råby regarded separately seemed to show the same trend, the results were not signi¤cant. There was no difference in survival between plants from different crossing categories for the Benestad plants.

Morphology The progeny in the crosses were signi¤- cantly taller (p<0.0001) at Benestad than at Grödby and Råby (Fig.10) as op- posed to the plants growing at their original ¤eld-sites (Table 1). There was no Fig. 9. Survival of Tephroseris integrifolia for signi¤cant difference between the Gröd- different crossing categories (cf. Fig. 7 & 8) by and the Råby plants (p=0.50). The (a) Survival of cohort 2003 in cultivation other morphological traits did not show 2005. any signi¤cant differences within or be- (b) Survival during the first winter 2004/ 2005 of cohort 2004. tween populations, except for a differ- (c) Survival during the winter 2004/2005 of ence between Grödby and Råby in width cohort 2003 in the field. of ¥ower head disc (data not shown). We

39 Inbreeding depression cannot explain the rapid decline

found a signi¤cant difference in height established in 1999 (Widén & Wetterin between crossing-categories within Be- 1999). The action plan comprised rein- nestad (p=0.023), where the plants from forcement of extant natural populations between sibs crosses were the shortest by transplanting adult plants raised and plants from between patch crosses from seeds in cultivation back to their were the tallest (Fig.10). There were no site of origin. These arti¤cial populations signi¤cant relationships between cros- are in most cases well separated from ex- sing-category and height for the Grödby tant natural stands reported in this (p=0.24) or the Råby (p=0.14) popula- paper (Widén, unpublished). A series of tions. For the rest of the morphological studies related to the conservation of T. traits we found no signi¤cant relation- integrifolia has been published (e.g. Wi- ships (data not shown). dén 1987, 1991, 1993, Widén & Anders- son 1993). Discussion Our main purpose of this study was to Most Swedish T. integrifolia populations explore the role of inbreeding depression have disappeared since the 19th century. in the decline of T. integrifolia by experi- The remaining populations have de- mental studies of progeny ¤tness in rela- creased in size since 1980. The decline in tion to cross proximity (cf. Byers 1998). number of populations represented in Inbreeding depression is often expressed the herbarium material towards the end at late life history stages (e.g. Karron of the 19th century cannot be explained 1989, Belaoussoff & Shore 1995, Nason by a decreased tendency to collect plant & Ellstrand 1995). Widén (1993) found material, since many specimens were clear indications of inbreeding depress- still collected, though from fewer popula- ion at early life history stages of our tions. Therefore, it is probable that most study species. The inbreeding experi- populations disappeared during the last ments performed on material derived quarter of the 19th century. Of the eight from the three populations Benestad, populations that remained when moni- Grödby, and Råby, showed a signi¤cant toring of T. integrifolia started in 1980, effect of crossing-category for most, but two (Lövhall and Åby) have disappeared not all, life-history stages; germination, altogether, and two others (Benestad and survival, and ¥owering at Grödby, and to Kverrestad) are on the verge of extinc- a certain extent at Benestad and Råby. tion today (2009). Height differed signi¤cantly between The main reason for the decline in crossing-categories for Benestad, but not number of populations of T. integrifolia for Grödby and Råby. For other morpho- during the 19th and 20th centuries is logical traits we found no signi¤cant dif- habitat loss caused by changing land ferences between crossing-categories at use. Nilsson & Gustafsson (1977) listed a any site. Several studies report a posi- range of known causes of extinction for tive effect on progeny vigour with in- individual populations; establishment of creasing crossing distances (e.g. Morán- arable ¤elds or cessation of grazing in Palma & Snow 1997, Byers 1998). semi-natural grasslands, forest planta- The magnitude of inbreeding depress- tion, abandonment of limestone quarries ion does not explain the dramatic de- and heavy fertilising combined with increase in population sizes since 1980. tensi¤ed grazing by cattle. Most of the experimentally induced in- The decline of T. integrifolia was ob- breeding depression can be referred to served by botanists in the early 20th cen- breeding between plants with the same tury (e.g. Andersson 1944), and despite maternal ¤eld parent, while plants with legal and conservation measures to pro- different maternal ¤eld parents from the tect the species the decline has con- same patch rarely gave lower ¤tness tinued; the species was protected by law than crosses between plants in different already in 1935 and an action plan was patches (cf. Oostermeijer, Altenburg, &

40 Inbreeding depression cannot explain the rapid decline

den Nijs 1995). Genetic depletion seems April–early May) and goes on through- not to be a threat to T. integrifolia, since out the season. Consequently most an earlier study showed signi¤cant ¥owering stems were probably grazed quantitative genetic variation for a before seed dispersal during the 1980s range of life-history and morphological (few estimations of the impact of grazing traits in both the Benestad and Grödby were made after 1982). The decline in populations (Widén & Andersson 1993). population sizes in the 1990s may partly The only morphological effect of be a result of the intense and unre- inbreeding that we found in our stricted grazing promoted by the NOLA experiments was a decrease in plant project during the 1980s. Although T. in- height. If inbreeding-depression indeed tegrifolia, being a poor competitor, seems played a role in the historical extinctions dependent on grazing to prevail, grazing of populations, we would have expected before midsummer could seriously ham- plant-height to decrease before popula- per seed-dispersal, since ¥owering stems tions went extinct. The general increase are readily eaten by the two main in plant height from 1818 till 1960, in- grazers, rabbit (Oryctolagus cuniculus) dicated by the herbarium specimens, is and cattle (Bos taurus). The action plan rather an indication of light shortage, focused on the grazing regime (Widén & possibly as a consequence of pastures Wetterin 1999), and there have been being overgrown – etiolation as a re- changes since 2000; grazing is restricted sponse to light shortage has previously before midsummer, which makes seed- been con¤rmed for T. integrifolia (Widén dispersal possible. However, the popula- & Andersson 1993). tions have continued to decrease in size The processes behind the rapid decline after 2000. Recruitment seems to have of T. integrifolia at the local population been very low during the last decades level are not fully understood, but the ex- contrary to the situation during 1980– periences gained during the period 1980 1983 (Widén 1987). to 2009 indicate that three mechanisms The extreme decrease in the formerly may be involved: large Tosteberga population could partly 1. Grazing of ¥owering stems – which be attributed to rabbit grazing. An ex- affects seed-dispersal. clusion experiment in 2004 at Toste- 2. Herbivory by molluscs – which af- berga revealed that rabbits were the fects seedling-survival. main herbivores at that site when cattle 3. Longer periods of drought – which were excluded (Widén, unpublished). mainly affects seedling-survival, Hunting of rabbits in this area was ex- but also ¥owering, seed-develop- tensive until the early 1980s, after which ment, and, when extreme, the sur- time it almost ceased, giving rise to a vival of adult, well-established rapid increase in the population size of plants. rabbits (Widén, personal observations). Few estimations of the proportion of re- A demographic study of the species 1980 productively successful ¥owering stems to 1983 revealed that grazing reduced at Tosteberga were made after 1982, but seed production, e.g. by 100% at Bene- grazing by rabbits as well as by cattle stad in 1981. The impact of grazing has probably inhibited most of the seed varied between zero and 97.5% at other production at this site since the early sites during 1980–1982 (Widén 1987). 1980s. Grazing was intense even before Grazing by livestock has been economi- 1980; Nilsson & Gustafsson (1977) cally supported since the 1980s, and has reported only about 50 ¥owering stems been an important management practice at Tosteberga in 1976. The rabbit was (the NOLA-project). However, grazing is successfully introduced for hunting in generally not restricted to any speci¤c Skåne in the early 20th century (Görans- time of the year, but starts early (late son, Frylestam, & Berg 1983), thus

41 Inbreeding depression cannot explain the rapid decline

grazing by rabbits cannot be responsible 2008 M. Magnusson (unpublished) found for the pronounced decline in number of that the slug A. lusitanicus preferred populations of T. integrifolia in the late both adult and seedling plants of T. in- 19th century. Further, rabbit grazing tegrifolia to three other grassland cannot explain the decline in population species (Pulsatilla vulgaris, Helianth- sizes after 1980 at other sites than Toste- emum nummularium, and Leontodon berga, because very little grazing on T. hispidus). Experimental sowing of seeds integrifolia by rabbits have been ob- of T. integrifolia at Benestad during the served elsewhere (Widén, personal ob- last ¤ve years has not led to any recruit- servations). Cattle grazing could have ment (Widén, unpublished). The seed- contributed to the decline in population lings, if ever observed, were often re- size at Benestad, Grödby S, Benestad, corded as eaten by molluscs, probably by Lövhall, and Åby. At Kverrestad, on the the invasive species A. lusitanicus, a other hand, grazing by cattle had little species common at the site, especially in effect on seed production, but tall vegeta- the wet year 2007. tion hampered seedling establishment A number of serious droughts during (Widén 1987). the 1990s and later made seedling estab- Generally, the seedling stage repre- lishment practically impossible, and also sents the most vulnerable part of the life killed many adult plants (Widén, unpub- cycle in plants (Harper 1977, Fenner & lished). Drought is one other major cause Thompson 2005) and herbivory, e.g. mol- of seedling mortality in plants (Cook lusc grazing, is one of the major causes of 1979, Mole & Westoby 2004). The demo- seedling deaths (Cook 1979, Mole & graphic study 1980–83 of T. integrifolia Westoby 2004). Many plants eaten by revealed a high recruitment rate, if seed molluscs in the autumn were recorded in production was not prohibited by grazing the demographic study of T. integrifolia of ¥owering stems or if seed germination 1980–82 (Widén 1987). After the wet was not repressed by a dense grass summers of 1980 and 1981 a high pro- sward (Widén 1987). However, even the portion of the plants had clear signs of short period of low precipitation in the damage caused by molluscs in the au- summer of 1982 killed young plants in tumn. July 1982 was dry and less favour- the permanent plots (Widén 1987). Mor- able for snails and slugs. Consequently, tality was clearly age-dependent except the damage caused by molluscs in the at the Benestad site, which has a dif- autumn 1982 was only half or even less ferent hydrology than the other sites of that recorded in the two previous (Regnéll 1976), leading to a lower mor- years (Widén 1987). The same trend with tality during the 1982 drought in Skåne. a high frequency of plants damaged by A severe drought occurred again in July– molluscs after a wet summer and lower August 1983 (Widén 1987), but unfortu- frequency of mollusc damage on leaves of nately no data was recorded in the per- T. integrifolia after dry periods has been manent plots after July 1983. observed in the arti¤cial populations at The rapidly decreasing viability of Benestad and Kverrestad in 2006–2008 seeds con¤rms that T. integrifolia popu- (Widén, unpublished). Several molluscs lations cannot rely on a seed-bank for have been observed on or close to leaves their persistence. However, while viabil- of T. integrifolia (Arion lusitanicus, ity decreased more in the inbred seeds, Cepaea hortensis, Deroceras reticulatum, they also took longer to germinate (cf. and Helix pomatia), although actual Widén 1993). Therefore, though we see eating has not been observed (Widén, un- no serious signs of inbreeding depression published). A. lusitanicus, e.g. is known as a result of within-patch crosses, we to be a threat to rare plants (Bruelheide can draw the conclusion that further in- & Scheidel 1999). breeding would make the species more In an experimental cultivation study vulnerable to changes in precipitation

42 Inbreeding depression cannot explain the rapid decline

during the germination phase. population has covered a much larger The evidence so far indicates little or area, but we don’t know it’s exact distri- no recruitment in the extant popula- bution. After the quarrying activities tions. The populations seem to consist of ceased at Benestad towards the second old individuals ¥owering with an inter- half of the 19th century (Regnéll 1976), val of several years. Less than 50% of the the patchy structure has probably been plants that ¥owered during 1980 to 1983 rather stable, since the habitat between did so more than once (Widén 1987), and the patches is unsuitable for T. integri- a high proportion of the plants present in folia. And because of the relative isola- permanent plots in 1980 and still alive in tion of the plants, it can be expected that 1983 (seedlings excluded) did not ¥ower the population would have lost some at all during this period (69% at Bene- genetic variance through genetic drift stad, 72% at Kverrestad, 43% at Toste- (cf. Leimu et al. 2006). The fragmenta- berga, and 47% at Grödby). Variation in tion and population size decrease at number of ¥owering stems of T. integri- Grödby seems to have been much more folia among years is regulated by a com- drastic, after a large part of the area was plex interaction of wet and dry summers. planted with pine (Pinus sylvestris) Drought is a major factor regulating the around 1960 (Nilsson & Gustafsson ¥owering frequency. The number of ¥ow- 1977). However, as the isolation of the ering stems decreases dramatically after present-living plants has only gone on one or several successive years with per- since the 1960s, one might expect that iods of drought. One case in point is the less genetic variation would have been decrease in number of ¥owering stems in lost than at Benestad. However, con- 2003, the year after a severe drought in sidering the shorter generation time at the late summer of 2002. The most dra- Grödby (according to Widén 1987), the matic decrease in number of ¥owering loss of genetic variation might be much stems occurred at Tosteberga in the early more rapid there than at Benestad. The 1990s. In 1992, drought even killed genetic distribution at Råby seems many adult plants (Widén, personal ob- confusing, since between-patch crosses servations). sometimes gave lower ¤tness than with- The decline in population size is also in-patch crosses. Since the between- manifested in areas occupied by the patch crossing distances are short com- plant. Many patches with T. integrifolia pared to the distances of the same cros- became extinct during the 1980s and sing category for the other populations, 1990s, both at Benestad and at Grödby and since the Råby patches are relatively (Fig.1). Patches that were small and iso- large and close to each other, it is hard to lated at the beginning of the study are imagine our results being a case of opti- probably remains of larger continuous mal crossing-distance (sensu Waser & areas covered by the plant. The larger Price 1989). Possibly, an extreme stirring patches occupied by T. integrifolia at and moving of soil as a consequence of Grödby are at present undergoing a frag- ploughing disrupted any kind of struc- mentation process. ture in the Råby population. We know the Benestad and the Grödby populations both to have decreased in Summary size since the 1850s, but in different The main reason for the pronounced his- ways and for different reasons. At Bene- toric decline of T. integrifolia in Sweden stad scattered plants found down-hill in is habitat loss. The recent extinction and the 1980s, close to the present occupied decline of population sizes are probably areas, as well as the numerous herba- caused by a combination of reduced seed rium specimens collected during the 19th production due to grazing, periods of century and till the species was pro- drought and mollusc herbivory on seed- tected by law in 1935, indicate that the lings. Today the extant populations seem

43 Inbreeding depression cannot explain the rapid decline

to consist of old individuals that ¥ower in Plant Population Biology. New York: with an interval of several years. In this Columbia University Press, 206–231. study, we found no clear signs of in- Cramer VA, Hobbs RJ. 2002. Ecological con- breeding depression that could explain sequences of altered hydrological regimes in fragmented ecosystems in southern the rapid decline in population sizes and Australia: Impacts and possible man- population number. Without proper agement responses. Australian Ecology management actions it is most likely 27:546–564. that the Swedish populations of T. integ- Fenner M, Thompson K. 2005. The ecology of rifolia will go extinct in a near future, seeds. Cambridge: Cambridge University and though inbreeding depression might Press. speed up the process, it does not play the Fischer M, Stöcklin J. 1997. Local extinctions leading part. of plants in remnants of extensively used calcareous grassland 1950–1985. Conserva- tion Biology 11:727–737. Acknowledgments Gaudeul M, Till-Bottraud I. 2008. Genetic Financial support to BW was provided by structure of the endangered perennial plant FORMAS (grant no 21.5/2002-0115) and the Eryngium alpinum (Apiaceae) in an alpine Swedish nature conservation agency (SNV) valley. Biological Journal of the Linnean and the county administrative board of Skåne Society 93:667–677. (Länsstyrelsen i Skåne län). We thank the Groppe K, Steinger T, Schmid B, Baur B, volunteer group (Floravårdsgruppen i Kris- Boller T. 2001. Effects of habitat tianstad) for assistance with annual census of fragmentation on choke disease (Epichloe natural populations in NW Skåne since 1999. bromicola) in the grass Bromus erectus. We thank Jane Jönsson and Eman Soubani Journal of Ecology 89:247–255. for their assistance in the green-house. We Göransson G, Frylestam B, Berg M. 1983. thank Stefan Andersson and Stig Isaksson for Fältviltet – 1. Biologi och miljö – hot och valuable comments on earlier drafts. skydd. Stockholm: Svenska jägareförbun- det. Harper JL. 1977. Population Biology of References Plants. London: Academic Press. Andersson O. 1944. Bidrag till Skånes Flora Karron JD. 1989. Breeding systems and 30: Senecio integrifolia. Botaniska Notiser levels of inbreeding depression in geo- 1944:445–458. graphically restricted and widespread Belaoussoff S, Shore JS.1995. Floral corre- species of Astragalus (Fabaceae). American lates and ¤tness consequences of mating- Journal of Botany 76:331–340. system variation in Turnera ulmifolia. Evo- Kiviniemi K, Eriksson O. 2002. Size-related lution 49:545–556. deterioration of semi-natural grassland Bruelheide H, Scheidel U. 1999. Slug herbi- fragments in Sweden. Divers. Distrb. 8:21– vory as a limiting factor for the geographi- 29. cal range of Arnica montana. Journal of Lawrence HA, Tayor GA, Millar CD, Lambert Ecology 87:839–848. DM. 2008. High mitochondrial and nuclear Byers DL. 1998. Effect of cross proximity and genetic diversity in one of the world’s most progeny ¤tness in a rare and a common endangered seabirds, the Catham Island species of Eupatorium (Asteraceae). Ameri- Taiko (Pterodroma magentae). Conserva- can Journal of Botany 85:644–653. tion genetics 9:1293–1301. Camargo JLC, Kapos V. 1995. Complex edge Leimu R, Mutikainen P, Koricheva J, Fischer effects on soil moisture and microclimate in M. 2006. How general are positive relation- central Amazonian forest. Journal of Tropi- ships between plant population size, ¤t- cal Ecology 11:205–221. ness, and genetic variation? Journal of Eco- Cillers SS, Williams NSG, Barnard FJ. 2008. logy 94:942–952. Patterns of exotic plant invasions in frag- Liu W, Dong M, Song Z, Wei W. 2009. Genetic mented urban and rural grasslands across diversity pattern of Stipa purpurea popu- continents. Landscape Ecology 23:1243– lations in the hinterland of Qinghai-Tibet 1256. Plateau. Annals of Applied Biology 154:57– Cook RE. 1979. Patterns of juvenile mortality 65. and recruitment in plants. In: Solbrig OT, Luijten SH, Dierick A, Gerard J, Oostermeijer Jain S, Johnson GB, Raven PH, eds. Topics B, Raijmann LEL, Den Nijs HCM. 2000.

44 Inbreeding depression cannot explain the rapid decline

Population size, genetic variation, and K-A. et al. Floran i Skåne. Vegetation och reproductive success in a rapidly declining, ut¥yktsmål. Lund: Lunds Botaniska Före- self-incompatible perennial (Arnica monta- ning. na) in The Netherlands. Conservation Bio- Shah A, Li D, Gao L, Li H, Möller M. 2008. logy 14:1776–1787. Genetic diversity within and among popula- Mable BK, Adam A. 2007. Patterns of genetic tions of the endangered species Taxus diversity in outcrossing and sel¤ng popula- fuana (Taxaceae) from Pakistan and impli- tions of Arabidopsis lyrata. Molecular Eco- cations for its conservation. Biochemical logy 16:3565–3580. Systematics and Ecology 36:183–193. Marquardt PE, Echt CS, Epperson BK, Smith UK. 1979. Biological ¥ora of the British Pubanz DM. 2007. Genetic structure, diver- Isles, Senecio integrifolius (L.) Clairv. sity and inbreeding of eastern white pine Journal of Ecology 67:1109–1124. under different management conditions. Snaydon RW, Davies MS.1976. Rapid Canadian Journal of Forest Research population differentiation in a mosaic envi- 37:2652–2662. ronment. IV. Populations of Anthoxanthum Matthies D, Bräuer I, Maibom W, Tscharntke odoratum at sharp boundaries. Heredity T. 2004. Population size and the risk of local 37:9–25. extinction: empirical evidence from rare Tyler T, Olsson K-A, Johansson H, Sonesson plants. Oikos 105:481–488. M. 2007. Floran i Skåne. Arterna och deras Mbora DNM, McPeek MA. 2009. Host density utbredning. Lund: Lunds Botaniska Före- and human activities mediate increased pa- ning. rasite prevalence and richness in primates Vandepitte K, Jacquemyn H, Roldán-Ruiz I, threatened by habitat loss and fragmenta- Honnay O. 2007. Landscape genetics of the tion. Journal of Animal Ecology 78:210– self-compatible forest herb Geum urbanum: 218. effects of habitat age, fragmentation and Mole AT, Westoby M. 2004. What do seedlings local environment. Molecular Ecology die from, and what are the implications for 16:4174–4179. evolution of seed size? Oikos 106:193–199. Van Geert A, Van Rossum F, Triest L. 2008. Morán-Palma P, Snow AA. 1997. The effect of Genetic diversity in adult and seedling interplant distance in mating success in populations of Primula vulgaris in a frag- federally threatened, self-incompatible Hy- mented agricultural landscape. Conserva- menoxys herbacea=H. acualis var. glabra tion Genetics 9:845–853. (Asteraceae). American Journal of Botany Van Rossum F, Echchgadda G, Szabadi I, 84:233–238. Triest L. 2002. Commonness and long-term Münzbergová Z. 2006. Effect of Population survival in fragmented habitats: Primula Size on the Prospect of Species Survival. elatior as a study case. Conservation Bio- Folia Geobotanica 41:137–150. logy 16:1286–1295. Nason JD, Ellstrand NC.1995. Lifetime Waser NM, Price MV. 1989. Optimal outcros- estimates of biparental inbreeding depres- sing in Ipomopsis: seed set and offspring sion in the self-incompatible annual plant ¤tness. Evolution 43:1097–1109. Raphanus sativus. Evolution 49:307–316. Widén B. 1987. Population biology of Senecio Nilsson Ö, Gusafsson L-Å. 1977. Projekt Lin- integrifolius (Compositae), a rare plant in né rapporterar 49–63. Svensk Botanisk Tid- Sweden. Nordic Journal of Botany 7:686– skrift 71:205–224. 704. Oostermeijer JGB, Altenburg RGM, den Nijs Widén B. 1991. Phenotypic selection on JCM. 1995. Effects of outcrossing distance ¥owering phenology in Senecio integri- and sel¤ng on ¤tness components in the folius, a perennial herb. Oikos 61:205–215. rare Gentiana pneumonanthe (Gentiana- Widén B. 1993. Demographic and genetic ef- ceae). Acta Bot. Neerl. 44:257–268. fects on reproduction as related to popula- Ouborg NJ. 1993. Isolation, population size tion size in a rare, perennial herb, Senecio and extinction: the classical and metapopu- integrifolius (Asteraceae). Biological Jour- lation approaches applied to vascular nal of the Linnean Society 50:179–195. plants along the Dutch Rhine-system. Oi- Widén B, Andersson S.1993. Quantitative kos 66:298–308. genetics of life-history and morphology in a Regnéll G. 1976. Naturreservatet Benestads rare plant, Senecio integrifolius. Heredity backar – vegetation och ¥ora. Svensk Bota- 70:503–514. nisk Tidskrift 70:17–42. Widén B, Wetterin M. 1999. Åtgärdsprogram Regnéll G. 2003. Benestads backar, In Olsson för bevarande av fältnocka (Tephroseris in-

45 Inbreeding depression cannot explain the rapid decline

tegrifolia). Stockholm: Naturvårdsverket. in south-west Western Australia. Biodiver- Willi Y, Van Buskirk J, Schmid B, Fischer M. sity and Conservation 1379–1395. 2006. Genetic isolation of fragmented popu- Yoshioka Y, Honjo M, Iwata H, Ninomiya S, lations is exacerbated by drift and selection. Ohsawa R. 2007. Pattern of geographical Journal of Evolutionary Biology 20:534– variation in petal shape in wild populations 542. of Primula sieboldii E. Morren. Plant Williams D A, Wang Y, Borchetta M, Gaines Species Biology 22:87–93. MS.2007. Genetic diversity and spatial Young A, Boyle T, Brown T. 1996. The popula- structure of a keystone species in fragmen- tion genetic consequences of habitat frag- ted pine rockland habitat. Biological Con- mentation for plants. TREE 11:413–418. servation 138:256–268. Young AG, Brown AHD, Zich FA. 1999. Gen- Yates CJ, Coates DJ, Elliott C, Byrne M. 2007. etic Structure of Fragmented Populations Composition of the pollinator community, of the Endangered Daisy Rutidosis lep- pollination and the mating system for a torrhynchoides. Conservation Biology shrub in fragments of species rich kwogan 13:256–265.

46 II Self-incompatibility does not explain declining population sizes

48 Self-incompatibility does not explain declining population sizes

Self-incompatibility does not explain declining population sizes of Tephroseris integrifolia (Asteraceae) Kerstin Isaksson, Department of Ecology, Section of Plant Ecology and Systematics, Lund University, Sweden

Abstract Small populations face problems of inbreeding depression and extinctions due to stochastic events. One of the problems that a small plant population might be challenged with is a reduced seed-set. This is often attributed to pollination failure, but could, in self-incompatible species, be an effect of loss of S-alleles through drift. In the rare and self-incompatible Tephroseris integrifolia (L.) Holub, seed-set has been found to be density-dependent in cultivation, and positively correlated with population size in wild populations. In this study of three populations in different stages of fragmentation I investigate the connection between self-incompatibility and reduced seed-set in fragmented populations, and whether this could be one of the causes of the severe decline of T. integrifolia in Sweden during the past decades. The main conclusions are: 1. Self-incompatibility does not prevent the plants from reproducing except in extremely small population fragments (with 1–3 plants). 2. There probably haven’t been many generation shifts since the populations were fragmented, since otherwise, a substantial number of S-alleles would have been lost through drift, and seed-set would have been much lower. 3. Considering how rapidly the population sizes are decreasing, these populations are likely to be extinct before SI would cause any reproduction dif¤culties.

Keywords: Sporophytic self-incompatibility, small populations, fragmentation, extinction, relict populations, threatened species.

Introduction giotti 2003), and an SI system is one ef¤- Self-incompatibility (SI) prevents plants cient mechanism to prevent inbreeding. from fertilising themselves and reduces However, when the supply of suitable the risk of being fertilised by close rela- mates is scarce, self-fertilisation or fer- tives (De Nettancourt 2001), which pro- tilisation of close relatives could be the tects them from inbreeding depression only option of sexual reproduction, and (Castric & Vekemans 2004). About 70% therefore an SI system would be a bur- of all angiosperm species have some form den (Demauro 1993). of SI system, and the evolution of SI is Low seed-set is frequently observed in thought to be part of the reason why small plant populations, and is usually angiosperms have been so successful attributed to low levels of pollen deposi- (Franklin-Tong & Franklin 2003). tion, as a result of infrequent pollinator There is a high risk of inbreeding visits (Widén 1993, Aizen & Feinsinger depression in small populations (Gag- 1994, Byers 1995, Ågren 1996), or to lack

49 Self-incompatibility does not explain declining population sizes

of compatible pollen as a result of low planation could be the lack of dominance variation in self-incompatibility alleles relationships, which would increase the (S-alleles) (Byers 1995, Wagenius, Lons- number of compatible crosses, and there- dorf, & Neuhauser 2007). Some attempts by the number of generations before ex- have been made to estimate the effects of tinction, but also speed up the loss of SI on the persistence of small popula- dominant alleles. Wagenius et al. (2007) tions by means of computer simulations documented a positive relationship be- (e.g. Byers & Meagher 1992, Wagenius, tween population size and seed-set, as Lonsdorf, & Neuhauser 2007, Goodwillie well as plant-density and seed-set in the 2008, Levin, Kelley, & Starkar 2009). self-incompatible Echinacea angustifolia The simulations take into account differ- (Asteraceae). In the same study, they ent aspects of SI, such as: incompatib- performed simulations based on the pro- ility at the gametophytic or sporophytic perties of the E. angustifolia populations stage, the number and frequencies of dif- they surveyed in the ¤eld. The simula- ferent S-alleles in the populations at the tions were performed both for a sporo- start of the simulation, and whether phytic SI system, with completely co- there are any dominance-relationships dominant alleles, and for a self-compat- between the S-alleles. The simulations ible system. They found that following also take into account different aspects fragmentation, populations of self-in- of population dynamics, such as produc- compatible plants, though they suffered tion and dispersal abilities of pollen and from less inbreeding depression than seeds, generation overlaps, inbreeding populations of self-compatible plants, de- depression, and mutation rates (both in creased more quickly. They also found S-alleles and in other alleles). Some of that the burden of SI becomes more evi- the questions asked are: how long will dent with time, and that the number of the population persist? and at what rate S-alleles decreases through drift. will S-alleles be lost? Byers & Meagher In their review on SI in natural popu- (1992) simulated populations of different lations, Castric & Vekemans (2004) more sizes and with different frequencies of S- or less dismiss SI as a threat to small or alleles that had strict hierarchic domi- fragmented populations. They hypoth- nance relationships, and found that at esise that fecundity selection would en- population sizes of under 50 individuals, sure the establishment of S-alleles that S-alleles will be lost relatively quickly enter a population through gene-¥ow (approximately one allele every three from other populations, or even from generations). Levin, Kelley, & Starkar other species, and conclude that SI (2009) simulated populations with either would only be a threat to a population gametophytic or sporophytic SI, as well under extreme isolation. as mendelian populations, and varying In small or fragmented populations of population sizes, initial numbers of S- self-incompatible plants seed-set is often alleles, and dispersal abilities of pollen low (Widén 1993, Heschel & Paige 1995, and seeds, while assuming complete co- Luijten et al. 2000, Brys et al. 2004), and dominance between alleles, and no in- sometimes completely absent (e.g. breeding depression. These simulations Demauro 1993). The self-incompatible showed that although small populations plant species Tephroseris integrifolia is of self-incompatible plants will quickly rare in Sweden, and has constantly de- move towards extinction, particularly if creased in both population number and SI is sporophytic, S-alleles will not be size since the 19th century, and despite lost till extinction is very close at hand. extensive conservation measures, it is They attribute the fact that S-alleles are still decreasing (Widén 1987, Widén & not lost in their model, to their assump- Wetterin 1999, Isaksson & Widén paper tion of equal frequencies of S-alleles at I). Studies of natural populations of T. the start of the simulation. Another ex- integrifolia have shown a strong positive

50 Self-incompatibility does not explain declining population sizes

relationship between seed-set and num- vince (Isaksson & Widén paper I). The ber of ¥owering stems in population or preferred habitat type is nutrient-poor, patch (Widén 1993, Isaksson & Widén calcareous grassland with little competi- paper I), and that isolated individuals – tion from other plants, and in Sweden it individuals which grow alone, or very is found mostly in heavily grazed pas- few together, more than 10 metres away tures (Widén 1987). from conspeci¤c plants – seldom set seed (Isaksson & Widén paper I). In cultiva- SI in Tephroseris integrifolia tion experiments of the same species, a All of the SI systems that have been positive relationship has been found be- studied in the Asteraceae family have tween plant density and seed-set (Widén been found to be sporophytic (e.g. Byers 1993). Since seed-set was not related to 1995, De Nettancourt 2001, Hiscock & population size after hand-pollinations, Tabah 2003, Kirchner, Robert, & Colas Widén (1993) suggested lack of pollina- 2006, Ferrer et al. 2009), including spe- tors as the most plausible explanation of cies of the genus Senecio (Brennan et al. low seed-set in small populations, al- 2003, Hiscock & Tabah 2003), to which though he didn’t rule out the possibility Tephroseris is closely related. It is, there- that a lack of S-alleles could cause a de- fore, reasonable to assume that the SI creased seed-set. system in T. integrifolia is sporophytic as While SI promotes genetic variation, well. In a sporophytic SI system, the SI- to function properly it also requires gen- phenotype of the pollen is determined etic variation in the S-locus, or else it not by the haploid genotype of the pollen, will hamper reproduction. Does SI in but by the diploid genotype of the pollen small populations really protect them donor. Sporophytic SI is a multi-allelic against inbreeding depression, or does it one-locus system, often with complicated simply accelerate the rapid wipe-out of hierarchical dominance relationships be- the whole population? In this study I in- tween alleles (De Nettancourt 2001). tend to study the relationship between Evidence for complicated hierarchical population size and seed-set following dominance has also been documented in different levels of inbreeding, to investi- Senecio (Hiscock & Tabah 2003). Some gate if SI could be contributing to the on- unexpected compatible crosses in Se- going decrease of T. integrifolia. necio have been attributed to a rudi- mentary gametophytic element in the Material and methods sporophytic SI mechanisms (Hiscock Study species 2000a & b, cf. Lewis, Verma, & Zuberi Tephroseris integrifolia (L.) Holub is a 1988). As a result of dominance-relation- self-incompatible perennial herb of the ships between alleles, reciprocal crosses family Asteraceae. The ¥ower heads (1– can be compatible in one direction and 4, sometimes more) are arranged in a co- incompatible in the other. Another char- rymb. It has 3–10 leaves arranged in a acteristic of sporophytic SI systems is rosette, and there are usually 1–3 small that they ‘leak’, so that even self-pollina- stem leaves. T. integrifolia ¥owers in tion often gives at least one seed, or so May to June, and it rarely ¥owers in two that crosses which ought to be incom- successive years. Most seeds germinate patible give seeds (Lewis 1994, Hiscock the ¤rst year, in late summer or autumn, 2000b). I regard cross-pollinations that and there is practically no seed-bank give no, or extremely low (see paragraph (Widén 1987). The distribution is dis- about Incompatibility below, and Fig. 1– junct with the westernmost populations 2), seed-set as incompatible, and all occurring in Britain and the easternmost other cross-pollinations as compatible. in Japan (Smith 1979). In Sweden there Variation in seed-set of putatively com- are six extant T. integrifolia populations patible crosses I analyse separately, and – all in Skåne, the southernmost pro- in the ¤nal section I discuss possible ex-

51 Self-incompatibility does not explain declining population sizes

planations of this variation, such as in- the north of the plantation (patches 4 breeding depression or variations in SI and 5). The distance between the two mechanisms. areas is about 400 metres.

Populations and sampling Råby Plants from three of the Swedish popula- The Råby population is situated about tions were used in this investigation – 300 metres north-west of the Grödby Benestad, Grödby, and Råby. The de- population (Isaksson & Widén paper I). scription of the populations is consistent Seeds were collected from three arti¤cial with what they looked like in 2000 and patches (that is, sections demarcated 2001, when the seeds for this investiga- within larger, continuous areas of tion were collected (for a more detailed plants) of about the same size as the Be- description of the populations below, see nestad patches, one from each of the Isaksson & Widén paper I). Since I am three larger patches. interested in how population structure affects local genetic structure, I used the The distance between the Råby popula- structure of the Benestad population, tion and the northern parts of the Gröd- which had the patchiest distribution, as by population is about 300 metres – a model for the seed collection in the shorter than the distance between the other two populations. northern and the southern part of the Grödby population. However, until re- Benestad cently (~1960, Nilsson & Gustafsson The Benestad population is situated 1977) the Grödby population was conti- within a nature reserve with grazed pas- nuous, and because I am more interested tures on calcareous soils. The topogra- in differences in population structure phy is a mosaic of wet and dry areas. T. than in genetic distances between the integrifolia grows in the dryer areas on a populations, I have chosen to regard south-west facing slope (Regnell 1976, Grödby and Råby as two separate popu- 2003), in nine rather well-de¤ned lations. patches, with 10 to 50 metres between each patch. The sizes of the patches are Cultivation and crossing experiments between a few and 50 m2, and in 2000 be- Seeds were sown in plastic pots in the tween 1 and 33 plants ¥owered in the green-house. Seedlings were transferred separate patches (Isaksson & Widén pa- to a common garden, and the plants were per I). Seeds were collected from three kept outside until just prior to ¥owering patches with a maximum distance of 10 in 2002 and 2003 when cross-pollina- metres between plants within a patch. tions were executed. During the time of cross-pollinations (April and May), Grödby plants were kept in an insect-free green- Until the end of the 1950s, when most of house environment with minimum risk the habitat was destroyed to make space of unwanted cross-pollinations. I per- for a pine plantation, this was one of the formed cross-pollinations by rubbing two largest known T. integrifolia populations ¥ower heads against each other on two in Sweden (Nilsson & Gustafsson 1977). different occasions at an interval of one Today plants grow around the pine-plan- to three days. The ¤rst cross-pollinations tation, mostly to the south and the north. were carried out when ray ¥orets and Seeds were colleted from ¤ve patches – outer tubular ¥orets were developed, and from three arti¤cial patches about the the second when all ¥orets in the two same size as the Benestad patches with- ¥ower heads were fully developed. Doing in the continuous area to the south of the just the latter cross-pollination would re- plantation (patches 1–3, see Isaksson & sult in a lower seed-set, since fully devel- Widén paper I), and from two patches to oped ¥orets have usually lost some of

52 Self-incompatibility does not explain declining population sizes

their pollen. The main set of cross-pollin- Cross-pollinations were evenly distri- ations was performed in 2002 on plants buted between the ¤rst three categories. grown from seeds collected in all three Since T. integrifolia, like other Aster- populations in 2000. Additional crosses aceae species, will eventually self-pol- were performed in 2003 on plants col- linate, if they don’t receive pollen from lected from the Råby population in 2000 other plants, no assistance is required and 2001. In total, 925 ¥ower heads were for. I have noted that rubbing ¥owering cross-pollinated, and most of the cross- plants with cotton-buds doesn’t lead to pollinations were reciprocal (some ¥ower higher seed-set than leaving the plants heads had to be removed from the in- to pollinate themselves – which is usual- vestigation, for example, if they had ly no seeds at all. In the following I will withered). 90 of these were crosses be- call these ¥ower heads un-pollinated. tween populations, which were used as a control, to get an estimate of maximum Seed-set seed-set, assuming maximum difference From each ¥ower head, I counted the in S-alleles. The number of ¥ower heads number of well-developed achenes as included in the analysis was 196 for the well as the number of empty achenes. I Benestad population, 191 for the Grödby measured seed-set in the ¥ower head as population, and 448 for the Råby popu- the number of well developed achenes lation. divided by the total number of achenes.239 un-pollinated ¥ower heads Crossing-types were analysed and used as a control. In Cross-pollinations were categorised into the data-set from the cross-pollinations, four crossing-types, according to the the large number of zeroes together with level of inbreeding: a fairly normal distribution of the rest of • Between siblings – crosses between the seed-sets made it impossible to plants with the same ¤eld mother transform this into one manageable data • Within patch – crosses between set, which is why it was divided into two plants with different ¤eld mothers data sets, one binomial and one continu- but from the same patch ous. However, the undivided data set is • Between patches – crosses between shown in diagrams. plants with ¤eld mothers from different patches in the same popu- Incompatibility lation Seed-set in un-pollinated ¥ower heads • Between populations – crosses be- was normally 0 (in 177 cases out of 239), tween plants with ¤eld mothers and seldom above 0.02 (in 32 cases) (see from different populations Fig.1). Seed-set in cross-pollinated plants showed a peak at 0 (incompatible crosses), but there was also a high amount of crosses between 0 and 0.02 (Fig.2). Based on this, and on the fact that sporophytic SI systems are known to ’leak’ (Lewis 1994, Hiscock 2000b), I regarded cross-pollination resulting in seed-set below 0.02 as incompatible, and the rest as compatible. Incompatibility was analysed with χ2-tests, and plants were pooled within populations. The Rå- by plants from the three different cross- pollination sets (plants from the 2000 Fig 1. Seed-set in un-pollinated flower ¤eld seed collection cross-pollinated heads. 2002 and 2003, and plants from the 2001

53 Self-incompatibility does not explain declining population sizes

Fig 3. Seed-set in all three populations to- Fig 2. Seed-set in pollinated flower heads. gether, unpollinated flower heads as well as between population crossed flower ¤eld seed collection cross-pollinated heads included. 2003) were analysed both together and separately. Results Seed-set in 835 cross-pollinated ¥ower Reciprocity heads (between populations crosses ex- I checked for reciprocity in 273 reciprocal cluded) varied between 0 and 0.98 crosses i.e. two ¥ower heads for each (Fig.2). In 62 of these no seeds were de- cross-pollination – if one cross had to be veloped at all, and in 43 ¥ower heads excluded, e.g. because the ¥ower head seed-set was higher than 0 but lower had withered before seeds could develop, than 0.02. When comparing all cross-pol- the reciprocal ¥ower head was excluded linations, regardless of population origin from the analysis as well. I divided seed- (Fig.3), sib-crosses had the lowest seed- set into three categories; seed-set=0, set among the cross-pollinated plants. 00.02, Within-patch crosses had a higher seed- and checked if the seed-set in each recip- set than sib-crosses but lower than rocal pair fell into the same category. between-patch crosses. Crosses between populations gave a seed-set similar to Seed-set of compatible crosses that of between-patch crosses. When Only compatible crosses (seed-set above populations are treated separately, the 0.02) were included in the analysis. To pattern is similar to that seen when achieve approximate normality, seed-set origin is not taken into account (Fig.4). was transformed with the logit function: However, the effect seems to be stronger logit(seedset)=ln(seedset/(1-seedset)) in the most fragmented population, Be- Seed-set was analysed for all three popu- nestad, and weaker in the least frag- lations together, and for the populations mented population, Råby. separately, using a general linear model. I tested for the effect of crossing-type, Incompatibility population, and paternal ¤eld grand- The incompatibility analysis revealed mother (nested within population). I also that the proportion of incompatible cros- tested for the interaction between crosses differed signi¤cantly between crossing-type and population, and the inter- sing categories in the Benestad (p= action between crossing-type and pater- 0.00073) and Grödby (p=0.0038) popula- nal ¤eld mother. tions, but not in the Råby population (p=0.279) – regardless of whether the All tests were performed in SPSS 16.0 Råby plants were analysed all together (SPSS Inc., Chicago, IL). or divided into groups according to when they were collected or pollinated (Fig.5).

54 Self-incompatibility does not explain declining population sizes

Fig 4. Seed-set in different crossing-types. Fig 5. Portion of compatible crosses (seed- Crosses which gave 0 seed-set included. set>0.02). Number of compatible crosses and total number of crosses above bars. Of the 90 between-populations crosses all except one (which produced only one seed) were compatible (not shown). Discussion Seed-set has been found to be related to Reciprocity population size in self-incompatible From the sample of 273 reciprocal cros- plants (Widén 1993, Heschel & Paige ses that were checked for reciprocity, of 1995, Luijten et al. 2000, Brys et al. the 49 crosses that had a seed-set of 0 2004). Widén (1993) also demonstrated (exactly) in at least one direction, 11 had that seed-set is density-dependent in T. a seed-set of 0 (exactly) in both direc- integrifolia. He concluded that the rea- tions, and 22 had a seed-set greater than son for the low seed-set in small popula- 0.02 in the other direction. Two recipro- tions is lack of pollen. In the ¤eld, T. in- cal crosses gave seed-set between 0 and tegrifolia plants growing more than 10 0.02 in both directions. metres away from other plants rarely set seed (Isaksson & Widén paper I). In an Seed-set of compatible crosses earlier study which includes the ¤eld The effect of crossing-category in com- parents of the plants in this study (Isaks- patible crosses (671 investigated ¥ower son & Widén paper I), seed-set in the heads) was the strongest in the Benestad ¤eld was positively correlated with the population, where all three categories number of ¥owering plants within a differed signi¤cantly from each other (p<0.001, Fig.6). The effect was the same in the Grödby population (p=0.003, Fig.6), but only sib-crosses and between-patch crosses differed signi¤cantly from each other. For the Råby population, though the tendency seemed to be the same, there was no signi¤cant effect of crossing-type on seed-set (p=0.158, Fig.6). Between populations crosses produced a lower seed-set than within population crosses from Benestad (p=0.032), but it was not signi¤cantly different from the seed-set of within population crosses in Grödby (p=0.341) (not shown). Råby plants were not used Fig 6. Seed-set in compatible crosses (seed- as fathers in the between-populations set<0.02 excluded). crosses.

55 Self-incompatibility does not explain declining population sizes

patch (where seed-set was measured set in the ¤eld (Isaksson & Widén paper both as number of incompatible crosses I). That this might be related to a lack of compared with number of compatible S-alleles is con¤rmed by the greater crosses (assuming that plants with a effect of crossing-type on seed-set in the seed-set lower than 0.02 had not received Benestad plants. However, in the cros- any compatible pollen) and as average sing experiments, seed-set of compatible seed-set). crosses was higher for within and be- I conclude that the SI system of T. in- tween patch crosses for the Benestad tegrifolia is sporophytic, based on two plants than for the others. It is, there- facts: 1. The reciprocal differences seen fore, possible that the lower seed-set in in the cross-pollinations, which indicate the Benestad population is caused by a dominance relationship between S-al- factors other than SI, such as less fre- leles (cf. Hiscock & Tabah 2003). 2. The quent pollinator visits, perhaps as a con- fact that some ¥ower heads which had sequence of the general scarcity of T. in- been left to self-pollinate set a few seeds, tegrifolia at this site. The between popu- which indicates a ’leak’ in the SI system lations crosses (Benestad–Grödby), gave not found in gametophytic systems (De a higher seed-set when the Benestad Nettancourt 2001). This is also consist- plants acted as mothers, than when the ent with several studies of Asteraceae Grödby plants acted as mothers. This species in which SI was found to be might be an example of purging of sporophytic (e.g. Demauro 1993, Byers genetic load, since the Benestad patches 1995, Kirchner, Robert, & Colas 2006, have been small and isolated for a long Abbott et al. 2009, Ferrer et al. 2009). A time. However, there is also a possibility dominance relationship between S-al- that the Benestad plants were best leles increases the amount of compatible suited for the green-house conditions – crosses (Bateman 1952), but it also in- in consistence with Crnokrak & Barrett creases the risk of S-alleles being lost, (2002), who suggest that cases of in- particularly since the most dominant al- creased ¤tness after generations of in- leles occur in lower frequencies than the breeding might be the result of selection rest. Similarly, transient SI (an SI sys- for laboratory conditions. Nevertheless, tem where plants become self-compat- though seed-set for the natural popula- ible with age) also leads to a higher tions is lower in Benestad than in Gröd- amount of compatible crosses, as well as by and Råby, seed-set in the Benestad a more rapid loss of S-alleles (Goodwillie patches was on average higher than in 2008). similar-sized patches in Grödby and The variations of seed-set of compat- Råby (Isaksson & Widén paper I), which ible crosses could perhaps be attributed speaks in favour of the purging hypo- to the ‘irregularities’ of sporophytic self- thesis. It is interesting that in Widéns incompatibility (De Nettancourt 2001, study from 1993, though Grödby was Hiscock & Tabah 2003). Another possible found to have a lower seed-set than Be- explanation of low seed-set in some com- nestad both after sib-crosses and after patible crosses could be inbreeding de- within population crosses (it is not stated pression – cross-pollination of genetic- whether within population crosses were ally similar individuals (though different performed within or between patches) at the S-locus) leading to homozygosity the difference in seed-set between these for deleterious recessive alleles (Wiens et two crossing-categories is greater for Be- al. 1987, Charlesworth 1989, Allphin, nestad than for Grödby! Wiens, & Harper 2002). SI protects plants against inbreeding, Of the three populations investigated and it decreases the chances of seed-set in the present study, it is Benestad which following cross-pollination between two had the smallest and most isolated closely related plants. Since population patches, and which had the lowest seed- fragmentation leads to a general de-

56 Self-incompatibility does not explain declining population sizes

crease in genetic variation as a result of S-alleles to reach zero. When comparing increased genetic drift, we might also ex- these results of Wagenius et al. (2007) pect a decrease in variation in S-alleles. with the results of my crossing-experi- Willi et al. (2005) in a study of popula- ments – where I found a high portion of tions of the self-incompatible Ranuncu- compatible crosses within fragments lus reptans, found that inbreeding de- (even between siblings, about 60% of the pression was high in small populations, crosses were compatible), it is reason- and concluded, therefore, that SI might able to assume that the number of gener- not protect small populations from in- ations since fragmentation is small. This breeding. The suggestions of Castric and means that the T. integrifolia popula- Vekemans (2004), that populations tions in Skåne are probably relicts of might be rescued from lack of S-alleles larger populations. If the populations re- through gene-¥ow from other popula- main small, but the plants manage to tions, or through occasional hybridisa- reproduce successfully, it is likely that S- tions with closely related species, hardly alleles will be lost through drift, and that applies to T. integrifolia in Skåne, since seed-set will decrease. However, if the the present populations grow very far populations continue to decrease in size from each other, and the only species of with the same speed as they have done the same genus, T. palustris is even more during the past three decades, they will uncommon, and grows in completely dif- probably have disappeared before drift ferent habitat types (Wigermo & Hå- could alter any gene frequencies. kansson 2005). It is dif¤cult to estimate the conse- Acknowledgements quences, either negative or positive, of Financial support to Björn Widén was pro- being self-incompatible, through the vided by FORMAS (grant no 21.5/2002-0115) study of real populations, or using com- and the Swedish nature conservation agency puter simulations. In simulations, it is (SNV) and the county administrative board of Skåne (Länsstyrelsen i Skåne län). I thank dif¤cult to take all necessary factors into Stefan Andersson and Ullrika Sahlin for stat- account (inbreeding depression, number istical advice. I thank Louise Hathaway, of S-alleles, gene ¥ow between popula- David Ståhlberg, and Björn Widén for valu- tions, population sizes, habitat fragment able comments on earlier drafts. sizes, generation times, and dispersal distances for seeds and pollen) and to References give them realistic values. In the study of Abbott RJ, Brennan AC, James JK, Forbes real populations, you get only a momen- DG, Hegarty MJ, Hiscock SJ. 2009. Recent tary image of a process, which would hybrid origin and invasion of the British have to continue for perhaps a hundred Isles by a self-incompatible species, Oxford ragwort (Senecio squalidus L., Asteraceae). years before you might see any change in Biol Invasions 11:1145–1158. SI variation as a result of a decrease in Ågren J. 1996. Population Size, Pollinator population size. The empirical-experi- Limitation, and Seed Set in the Self-Incom- mental-simulatory study of Wagenius et patible Herb Lythrum salicaria. Ecology al. (2007) shows the importance of the 77:1779–1790. age of a habitat-fragment. Their simu- Aizen MA, Feinsinger P. 1994. Forest Frag- lation of a population of self-incompat- mentation, Pollination, and Plant Repro- ible Echinacea angustifolia (a species duction in a Chaco Dry Forest, Argentina. Ecology 75:330–351. similar to T. integrifolia in longevity and Allphin L, Wiens D, Harper KT. 2002. The ¥owering-frequency) did not reach ex- relative effects of resources and genetics on tinction until several hundred years reproductive success in the rare Kachina after fragmentation (Wagenius, Lons- Daisy, Erigeron kachinensis (Asteraceae). dorf, & Neuhauser 2007), and even in the Int. J. Plant Sci. 163:599–612. smallest habitat fragments (100 m2) it Bateman AJ. 1952. Self-Incompatibility Sys- took 100 years for the effective number of tems in Angiosperms. Heredity 6:285–310.

57 Self-incompatibility does not explain declining population sizes

Brennan SC, Harris SA, Tabah DA, Hiscock Simulation Model and Empirical Test. Evol- SJ. 2003. The population genetics of sporo- ution 62:2105–2111. phytic self-incompatibility in Senecio Heschel MS, Paige KN. 1995. Inbreeding De- squalidus L. (Asteraceae) I: S allele diver- pression, Environmental Stress, and Popu- sity in a natural population. Philosophical lation Size Variation in Scarlet Gilia Transactions of the Royal Society in London (Ipomopsis aggregate). Conservation Bio- B 358:1047–1050. logy 9(1):126–133. Brys R, Jacquemyn H, Endels P, Van Rossum Hiscock SJ. 2000a. Self-incompatibility in F, Hermy M, Triest L, De Bruyn L, Blust Senecio squalidus L. (Asteraceae). Annals GDE. 2004. Reduced reproductive success of Botany 85 (Supp. A):181–190. in small populations of the self-incompat- Hiscock SJ. 2000b. Genetic control of self-in- ible Primula vulgaris. Journal of Ecology compatibility in Senecio squalidus L. 92:5–14. (Asteraceae): a successful colonizing spe- Byers DL. 1995. Pollen Quantity and Quality cies. Heredity 85:10–19. as Explanations for Low Seed Set in Small Hiscock SJ, Tabah DA. 2003. The Different Populations Exempli¤ed by Eupatorium mechanisms of sporophytic self-incompatib- (Asteraceae). American Journal of Botany ility. Philosophical Transactions of the 82:1000–1006. Royal Society of London B Biological Byers DL, Meagher TR. 1992. Mate avail- Sciences 358:1037–1045. ability in small populations of plant species Kirchner F, Robert A, Colas B. 2006. with homomorphic sporophytic self-incom- Modelling the dynamics of introduced popu- patibility. Heredity 68:353–359. lations in the narrow-endemic Centaurea Castric V, Vekemans X. 2004. Plant self- corymbosa: a demo-genetic integration. incompatibility in natural populations: a Journal of Applied Ecology 45:1011–1021. critical assessment of recent theoretical Levin DA, Kelley CD. Starkar S.2009. En- and empirical advances. Molecular Ecology hancement of Allee effects in plants due to 13:2873–2889. self-incompatibility alleles. Journal of Charlesworth D. 1989. Evolution of Low Ecology 97:518–527. Female Fertility in Plants: Pollen Limita- Lewis D. 1994. Gametophytic-sporophytic in- tion, Resource Allocation and Genetic Load. compatibility. In: Williams EG, Clarke AE, TREE 4:289–292. & Knox RB. (eds.) Genetic control of self- Crnokrak P, Barrett SCH. 2002. Purging the incompatibility and reproductive devel- genetic load: a review of the experimental opment in ¥owering plants. Kluwer Aca- evidence. Evolution 56:2347–2358. demic Publishers. De Nettancourt D. 2001. Incompatibility and Lewis D, Verma SC, Zuberi MI. 1988. Incongruity in Wild and Cultivated Plants. Gametophytic-sporophytic incompatibility Second, totally revised and enlarged edi- in the Cruciferae–Raphanus sativus. Her- tion. Springer-Verlag. edity 61:355–366. Demauro MM. 1993. Relationship of Breeding Luijten SH, Angelo K, Gerard J, Oostermeijer System to Rarity in the Lakeside Daisy B. 2000. Population Size, Genetic Variation (Hymenoxys acualis var. glabra). Conserva- and Reproductive Success in a Rapidly De- tion Biology 7:542–550. clining, Self-Incompatible Perennial (Arni- Ferrer MM, Good-Avila SV, Montaña C, ca montana) in The Netherlands. Conserva- Domínguez CA, Eguiarte LE. 2009. Effect tion Biology 14(6):1776–1778. of variation in self-incompatibility on pollen Nilsson Ö, Gustafsson L-Å. 1977. Projekt limitation and inbreeding depression in Linné rapporterar 49–63. Svensk Botanisk Flourensia cernua (Asteraceae) scrubs of Tidskrift 71:205–224. contrasting density. Annals of Botany 103: Regnéll G. 1976. Naturreservatet Benestads 1077–1089. backar – vegetation och ¥ora. Svensk Bota- Franklin-Tong N, Franklin CH. 2003. Gamet- nisk Tidskrift 70:17–42. ophytic self-incompatibility inhibits pollen Regnéll G. 2003. Benestads backar, In Olsson, tube growth using different mechanisms. K-A. et al. Floran i Skåne. Vegetation och TRENDS in Plant Science 8(12):598–605. ut¥yktsmål. Lund: Lunds Botaniska För- Gaggiotti OE. 2003. Genetic threats to ening. population persistence. Ann. Zool. Fennici Smith UK. 1979. Biological ¥ora of the British 40:155–168. Isles, Senecio integrifolius (L.) Clairv. Goodwillie C.2008. Transient SI and the Journal of Ecology 67:1109–1124. Dynamics of Self-Incompatibility Alleles: A Wagenius S, Lonsdorf E, Neuhauser C.2007.

58 Self-incompatibility does not explain declining population sizes

Patch Aging and the S-Allee Effect: Breed- tegrifolia). Stockholm: Naturvårdsverket. ing System Effects on the Demographic Re- Wiens D, Calvin CL, Wilson CA, Davern CI, sponse of Plants to Habitat Fragmentation. Frank D, Seavey SR. 1987. Reproductive The American Naturalist 169(3):383–397. success, spontaneous embryo abortion, and Widén B. 1987. Population biology of Senecio genetic load in ¥owering plants. Oecologia integrifolius (Compositae), a rare plant in 71:501–509. Sweden. Nordic Journal of botany 7:687– Wigermo C, Håkansson, C.2005. Åtgärdspro- 704. gram för bevarande av kärrnocka (Tephro- Widén B. 1993. Demographic and genetic ef- seris palustris). Naturvårdsverket. fects on reproduction as related to popula- Willi Y, Van Buskirk J, Fischer M. 2005. A tion size in a rare, perennial herb, Senecio Threefold Genetic Allee Effect: Population integrifolius (Asteraceae). Biological Jour- Size Affects Cross-Compatibility, Inbreed- nal of the Linnean Society 50:179–195. ing Depression and Drift Load in the Self- Widén B, Wetterin M. 1999. Åtgärdsprogram Incompatible Ranunculus reptans. Genetics för bevarande av fältnocka (Tephroseris in- 169:2255–2265.

III No connection between population size and local genetic variation

62 No connection between population size and local genetic variation

No connection between population size and local genetic variation in AFLP-markers in a threatened plant species (Tephroseris integrifolia)

Kerstin Isaksson, So¤e Nordström, and Björn Widén, Department of Ecology, Section Plant Ecology and Systematics, Lund University, Sweden.

Abstract Genetic population structure is the consequence of past dispersal and survival. Small-scale genetic variation in plant populations is essential for sexual reproduction, since plants to a great extent reproduce with their neighbours. As a consequence of extensive changes in the European cultural landscape, many plant populations have disappeared or become smaller or fragmented – processes which are likely to lead to genetic depletion. However, in nine Estonian and two Swedish populations of a rare plant species, Tephroseris integrifolia (L.) Holub (Asteraceae), we found no relation between population size and local (within ~10 m2) genetic variation in AFLP markers. One possible explanation for the lack of difference in local genetic variation is that the processes leading to a decrease in population sizes and numbers also have lead to a halted recruitment, thus we might now have relict populations which do not reproduce at all, and accordingly, what we see now is a more or less random sample of the populations as they were before land management changed. Analyses of genetic distances between populations revealed that the Swedish populations were signi¤cantly different from the Estonian populations, but also that one of the Estonian populations, from a relatively isolated island, was signi¤cantly different from the rest, and also seemed to harbour less genetic variation.

Key words: Local genetic variation, AFLP, rare species, small populations.

Introduction duction, dependent on what conspeci¤c The genetic structure within a plant plants are within ¥ying-distance of the population is created through a process pollinating insects (Kunin 1993, Wilcock of dispersal, establishment, and sur- & Neiland 2002). Habitat fragmentation vival. This process is dependent on sev- can decrease access to suitable pollen-do- eral factors. Since sexual reproduction in nors, which could lead to reduced seed- plants often takes place between individ- set through pollination failure (Wilcock uals which are relatively close to each & Neiland 2002, Tomimatsu & Ohara other, changes in pollinator behaviour, 2006), as well as to inbreeding depress- seed dispersal mechanisms, or plant sur- ion (Van Geert, Van Rossum, & Triest vival may change the chances of plant re- 2008). The genetic composition of a plant production. Insect pollinated, out-cros- population may give us some clues to sing plants are, for their sexual repro- past seed-dispersal mechanisms, as well

63 No connection between population size and local genetic variation

Table 1. Descriptions and geographic coordinates of populations for AFLP analysis. Gray indicates that individuals were not included in the analysis of genetic variation within population. For descriptions of primer combinations (B13, G20, & Y10), see Table 2.

Popluation – sample Population description Estonian populations Small population along gravel road surrounded by overgrown Allika pastures Allika 1

Allika 2 Small population within grazed pastures (number of flowering Järve stems might be under estimated due to recent grazing) Very small island in Estonian archipelago with dense stands of Kadakalaid Juniperus communis . Kadakalaid 1

Kadakalaid 2 Close to the north coast, some plants on the verge of chalk cliffs Keila-Joa eroding into the see Large population within pastures on the site of an iron age Muuksi linnamägi fortress. Muuksi linnamägi 1

Muuksi linnamägi 2

Muuksi linnamägi 3 Fairly large population in grassland which showed no signs of Pakri neem recent grazing Very small population within a pasture which showed no signs of Sooääre having been grazed that year Small population within a pasture which showed no signs of Türisalu having been grazed that year Fairly large population on an island to the east of Hiumaa. Most Vohilaid of the plants growing on chalk gravel a few metres from the see. Vohilaid 1

Vohilaid 2

Vohilaid 3

Vohilaid 4

Swedish populations Small, patchy population within a nature reserve with yearly Benestad grazing Grödby S Small population within grazed pastures.

64 No connection between population size and local genetic variation

Flowering Population Individuals analysed Coordinates Stems 2005 Surface B13 G20 Y10 Total

21 110 m 2 15 59 °°°23’31N 1 1 1 1 024 °°°23’24E 59 °°°23’31N 14 13 12 14 024 °°°23’24E 59 °°°23’34N 30 130 m 2 9 11 9 11 024 °°°22’58E 200 0.2 km 2 15 58 °°°59’00N 8 8 8 8 023 °°°00’13E 58 °°°59’16N 7 7 7 7 023 °°°00’12E 59 °°°24’06N 50 1150 m 2 10 8 9 10 024 °°°19’00E 5000 000 0.5 km 2 18 59 °°°30’44N 4 3 3 5 025 °°°31’26E 59 °°°30’43N 3 3 3 4 025 °°°31’28E 59 °°°30’41N 6 6 7 9 25 °°°31’32E 59 °°°23’23N 1260 ~1000 m 2 11 11 12 12 024 °°°02’37E 59 °°°22’35N 12 15 m 2 9 7 7 9 024 °°°21’32E 59 °°°24’55N 12 16.5 m 2 7 4 6 8 024 °°°18’52E 3500 2000 m 2 20 58 °°°55’30N 0 0 2 2 023 °°°01’39E 58 °°°55’29N 11 11 10 11 023 °°°01’41E 58 °°°55’29N 4 4 3 5 023 °°°01’42E 58 °°°55’29N 2 2 1 2 023 °°°01’42E

55 °°°31’36N 37 50 m 2 9 9 9 11 013 °°°54’14E 56 °°°05’26N 230 ~1000 m 2 11 10 11 11 014 °°°30’51E

65 No connection between population size and local genetic variation

Fig. 1. Genetic diversity for three primer combinations in Swedish and Estonian T. integrifolia populations ranked by sizes. (Number of flowering stems in brackets.) Stars: B13, crosses: G20, and bars: Y10 (see Table 1). as to the future chances of sexual repro- Landscape fragmentation and changes duction for individual plants within the in land management may have extensive population (Ouborg, Piquot, & Van Groe- effects on the ecology and genetics of na- nendael 1999, Van Geert, Van Rossum, & tural populations (Honnay & Jacquemyn Triest 2008). 2007, Aguilar et al. 2008). The changes The genetic variance within a popula- in European cultural landscape during tion is predicted to decrease when popu- the 20th century have entailed intensi¤- lation size decreases, and has been con- cation and landscape simpli¤cation ¤rmed e.g. in a meta-study by Leimu et (Stoate et al. 2001). In Skåne, southern al. (2006) on 46 studies covering 41 spe- Sweden, a marked decrease in pasture- cies, but also in studies of single species land and hay-¤elds occurred already dur- (e.g. Lammi, Siikamäki, & Mustajärvi ing the 19th century, when there was a 1999, Landergott et al. 2001, Kang, ¤ve-fold increase in cultivated surface Jiang, & Huang 2005, Coppi, Mengoni, & (Emanuelsson et al. 2002). This process Selvi 2008). There are mechanisms continued during the 20th century, when which are believed to hamper the detri- many pastures were also abandoned to mental effects of decreasing population be overgrown with bushes, while at the sizes, such as a non-random mating sys- same time many remaining pastures tem, which lessens the risk of inbreed- have been subject to an increased use of ing, and strong selection, which helps arti¤cial manure (Emanuelsson et al. preserve healthy gene-combinations. 2002). When unexpectedly high levels of within After 1950, much of the agricultural population variation have been found in land in Estonia was abandoned, and small populations, this is usually re- many pastures were overgrown (Kana, ferred to recent decreases in population Kull, & Otsus 2008). The total grassland size (e.g. Luijten et al. 2000, Mhemmed, area decreased with only about 3.5% Kamel, & Chedly 2008, Coppi, Mengoni, from 1950 to 2002, but only 11% of the & Sevi 2008, Kang, Jiang, & Huang grasslands in 2002 were grasslands also 2005), while low levels of genetic vari- in 1950, which has to a great extent de- ation within populations is usually as- creased genetic diversity of Estonian sociated with historical bottlenecks or grasslands (Kana, Kull, & Otsus 2008). long periods of small population sizes During the Soviet period, although Est- (e.g. Landergott et al. 2001, Coppi, onian agriculture, as was the case in all Mengoni, & Selvi 2008). of Europe, was characterized by large-

66 No connection between population size and local genetic variation

Table 2. Primer combinations with numbers and sizes of bands, Fst, Gst, and mantel tests.

B13 G20 Y10 average

MseI primer GATGAGTCCTGAGTAA CTA GATGAGTCCTGAGTAA CAT GATGAGTCCTGAGTAA CAC —

EcoRI primer GACTGCGTACCAATTC ACT GACTGCGTACCAATTC AGG GACTGCGTACCAATTC ACC —

Identified bands 292 186 229 —

Polymorphic bands 178 145 160 —

Min. band length 42.5 bp 46 bp 47 bp —

Max. band length 499 bp 297 bp 384 bp —

Fst 0.063 0.075 0.092 0.07677

Gst 0.0883 0.1106 0.1236 0.1075

Mantel test r=0.59 (p=0.011) r=0.64 (p=0.007) r=0.60 (p=0.002) r=0.61 scale farming, small-scale farming was threatened (Ingelög, Andersson, & also carried on in private plots of about Thernborg 1993) and the populations are 0.6 ha (or more) per person, utilized by both larger (Isaksson pers. obs.) and farmers otherwise working at collective more numerous (Kukk 2004) than the farms (Abrahams 1994), something Swedish populations. which might have favoured species that We have chosen to compare eleven were typical of the traditional cultural populations of varying sizes – two Swe- landscape in Europe. The Estonian part dish and nine Estonian – to see if there is of this study is concentrated to the two a difference in genetic variation between counties Harjumaa and Hiiumaa, where Swedish and Estonian populations, and 40–50% of the arable land was aban- between populations of different sizes. doned during the ¤rst years of the 1990s Speci¤cally, we want to answer the (Peterson & Aunap 1998). following questions: The focus of this study is the genetic 1. Is there a relationship between variation within pollination-distance of a population size and small-scale gen- plant species, Tephroseris integrifolia, in etic variation? populations of different sizes. T. integri- 2. Is there a difference between Swed- folia is a very rare species in Sweden, ish and Estonian populations in and it has decreased rapidly and drasti- small-scale genetic variation? cally during the past 140 years – c.85% 3. Is there a relationship between gen- of the known Swedish populations have etic and geographic distances be- disappeared, and the remaining six po- tween the populations? pulations are continually growing smaller (Isaksson & Widén paper I). Since T. Materials and methods integrifolia has continued to decline in Study species Sweden during the past decades, despite The Field ¥eawort (Tephroseris integri- farreaching conservation measures folia) is a self-incompatible, composite taken to save the remaining populations plant which has a disjunct distribution (Widén & Wetterin 1999) we suspect that throughout temperate regions of Eur- there might be other causes for the on- asia, from Britain to Japan. It grows on going decrease in population sizes – e.g. calcareous, nutrient-poor grasslands inbreeding depression as a consequence (Smith 1979) – a threatened habitat type of genetic drift, following previous popu- in most of Europe. In Sweden T. integri- lation decreases. In Estonia, on the other folia is associated with intense grazing hand, though T. integrifolia is not ex- (Widén 1987). Both Swedish and Est- ceedingly common, it is not considered onian populations of T. integrifolia are

67 No connection between population size and local genetic variation

edge-populations compared with the was carried out using the AFLP kit for main distribution area. Seed dispersal normal-sized plant genomes from Gibco mechanisms have not been studied for BRL (Life Technologies, Täby, Sweden). this species, but the fruit being an For restriction, ligation and pre-ampli¤- achene with a pappus it is, like many cation we followed the protocol from the other Asteraceae species, likely to be manufacturer, but reduced amounts of classi¤ed as wind-dispersed (Benvenuti reagents were used. 2007). Relatively closely related species For selective ampli¤cation we used (of the genus Senecio and other Aster- three primer combinations (Table 2). The aceae species) have been shown to dis- reaction volume (5 µL) of the selective perse about 2 metres at wind speeds of PCR reactions contained 3.9 µL ddH2O, 4.5 metres per second (Sheldon & Bur- 0.5 µL 10x reaction buffer (100 mM Tris- rows 1973), and since the achenes of T. HCl pH8.3, 500 mM KCl, 15 mM MgCl2), integrifolia are similar to Senecio 0.023 µL AmpliTaq Polymerase (5 units/ achenes in shape and size, as well as in µL; Applied Biosystems Stockholm, Swe- length and width of the pappi, it is likely den), and 0.4 µL template DNA (14 ng/ that wind does not disperse their seeds µL). Size-variable fragments were ampli- much farther than a few metres. Pollina- ¤ed by a touch-down procedure with an tors that have often been observed on T. initial 14 cycles of denaturing for 60 s at integrifolia are syrphids (hover¥ies) and 94°C, annealing for 60 s. at 65°C in the muscids (¥ies) (Widén pers. obs.) and ¤rst cycle, and then dropping 0.7°C plants more than 10 metres away from every cycle until the annealing tempera- other plants rarely set seeds (Isaksson & ture reached 56°C, and extension for 90 Widén paper I, Widén 1993). This species s. at 72°C; followed by 23 cycles of de- therefore seems to have limited dis- naturing for 30 s. at 94°C, annealing for persal abilities, both for seeds and for 30 s. at 56°C, and extension for 90 s. at pollen. 72°C; and ended by a ¤nal extension for 10 min at 72°C. Populations and sampling Seeds were collected in June and July AFLP-band analysis from two Swedish and nine Estonian In each selective primer pair, the EcoRI populations (see Table 1 for details). primer was Cy5-labelled, to make the Seven populations were from the Est- ampli¤ed fragments visible on the se- onian mainland and two from small is- quencer. The PCR product from each re- lands to the east of the island Hiiumaa – action was diluted in 20 mL formamide Vohilaid and Kadakalaid – of which Ka- and mixed with home-made internal size dakalaid was the most isolated. In each standards (PCR fragments of which the population, seeds from ~10 plants were size had been determined through com- collected within a 2×5 m2 surface. In parison with commercially available size some cases, seeds were also collected standards from Amersham Bioscience) to from other patches within the same enable size determination of the frag- population. We sowed the seeds and col- ments. For each 40 well gel that was run lected young, fresh leaves for DNA-ex- on the sequencer, two wells were re- traction. Only one offspring from each served for a Cy5-labelled 50–500 bp ex- ¤eld mother was included in the ana- ternal standard (Amersham Bio- lyses. science). Fragments were separated with ALF DNA extraction and molecular methods Express II DNA analyzer (GE Health- DNA was extracted from young, fresh care). The banding patterns were ana- leaves according to Lodhi et al. (1994). lysed manually with ALFwin Fragment Ampli¤ed fragment length polymor- Analyser 1.03.01 software (GE Health- phism (AFLP) analysis (Vos et al. 1995) care). We scanned the gels for band

68 No connection between population size and local genetic variation

lengths above 40 (the double size of the quin input ¤les and POPGENE32 input primers) to about 500 bp. Maximum ¤les were created with AFLPdat in R. band length for each primer was chosen according to the maximum band length Results of the majority of the samples (Table 2). The three primer combinations gave Only clear bands above 0.1% of full de- each between 186 and 292 bands (see tection with a symmetrical shape on the Table 2). B13 gave the highest number of curve-view screen were selected for ana- bands, 292. Y10 gave 229 bands. G20 lysis. Samples with less than 20 clear gave 186 bands. We have chosen to ana- bands were excluded from the analysis. lyse the results of the three primer-com- The population origins of the samples binations separately. Primer combina- were kept unknown during this process tions Y10 and G20 gave relatively simi- to avoid subjective estimations. Two in- lar results for genetic diversity, while dividuals were present on every gel to primer combination B13 differed some- serve as additional control. what from the others. The analyses of genetic variation with- Analyses of genetic variation in populations showed no relationship Since the number of bands and band re- between population size and amount of solution differed between primer combi- local genetic variation for either of the nations, they were analyzed separately. primer combinations, and there was no We analysed within population diversity difference between the Swedish and the with AFLPdat (Ehrich 2006) in R version Estonian populations in amount of local 2.7.0 (R Development Core Team 2008), genetic variation. The Kadakalaid popu- which calculates Nei’s gene diversity lation, which is the most isolated, had with the formula least genetic variation of all the popula- D=n/(n-1)×[1-freq(1)2+freq(0)2)] tions (see Fig.1). (assuming that the band frequency cor- Most of the genetic variation was with- responds to allele frequency); and esti- in the populations – the average (be- mates a con¤dence interval for the diver- tween the three primer combinations) sities from bootstrapping over markers Fst for all populations was 0.077. Aver- (we used 1000 bootstrap replicates). Be- age Gst was slightly higher: 0.11. The tween-populations diversity tests and isolated Estonian island population, Ka- Mantel tests were performed in Arlequin dakalaid, was signi¤cantly different 2.000 (Schneider, Roessli, & Excof¤er from all of the other populations for each 2000). A PCO analysis was applied to the of the three primer combinations. The corrected average pairwise differences differences between the mainland Est- from the Arelquin analysis, using onian populations, as well as the Vohi- NTSYSpc2.2 (Rohlf 2005). The Mantel laid population, were sometimes signi¤- test was performed in Arlequin, com- cant and sometimes not. In all of the paring an Fst distance matrix for the three PCOs based on corrected average populations with a geographic distance pairwise differences (calculated with matrix done in AFLPdat based on the Arlequin), the Benestad and Grödby geographic coordinates of the popula- populations were grouped together, the tions. Gst was calculated using POP- Kadakalaid population was isolated, and GENE (Yeh, Yang, & Boyle 1999). For the the rest of the Estonian populations were within population variation, only plants more or less grouped together (Fig.2a–c). collected within the 10 m2 plots were in- The two Swedish populations, Benestad cluded. For the Kadakalaid population, and Grödby, were never signi¤cantly dif- two such plots were analysed. For the be- ferent from each other, and both were tween-populations analysis, and the signi¤cantly different from the other Mantel test, plants collected outside the populations for the Y10 primer combina- 10 m2 plots were included as well. Arle- tion. For the G20 primer combination,

69 No connection between population size and local genetic variation

(a)

(b)

Fig. 2. PCOs applied to genetic average pairwise distances for (a) primercombination B13 (b) primercombination G20, and (c) primercombination Y10. Filled symbols for Estonian populations, open symbols for Swedish populations.

70 No connection between population size and local genetic variation

(c)

Grödby was not signi¤cantly different pected (Barrett & Kohn 1991, Ellstrand from Vohilaid. For the B13 primer combi- & Elam 1993), and has been con¤rmed in nation, Benestad was not signi¤cantly other studies, e.g. in a meta-study based different from Keila-Joa. The Mantel on 46 studies of 41 plant species by test showed a signi¤cant relationship be- Leimu et al. (2006), where self-incompat- tween genetic and geographic distances ibility seemed to enhance this relation- when all populations were included, but ship, but they found no difference be- not when only the Estonian populations tween common and rare species. How- were included. Due to the small number ever, there are a few studies which report of populations, no separate Mantel test unexpectedly high genetic variation in was performed for the Swedish popula- small populations. Globularia bisnagari- tions. ca is one example. It is a long-lived perennial which is probably dispersed Discussion through epizoochory (Honnay et al. The most striking result of this study is 2007). Arnica montana is another ex- that there is no relationship between ample. It is a long-lived, self-incompat- local genetic variation and population ible perennial with no seed-bank which size in the eleven investigated T. integri- has declined despite of habitat restora- folia populations. It is also worth men- tions (Luijten et al. 2000). The same phe- tioning that there is no difference in local nomenon is, perhaps not surprisingly, genetic variation between Estonian and found in tree-species (Soejima, Maik, & Swedish populations, although T. in- Ueda 1998, De Almeida Vieira & De tegrifolia is much more abundant in Carvalho 2008 Dzialuk et al. 2009), Estonia. where forest-fragments contain the same A positive relationship between popu- genetic variation as they did 50 years lation size and genetic variation is ex- ago, when they were parts of large for-

71 No connection between population size and local genetic variation

ests, where the remaining individuals very different genetic structure com- have more or less ceased to reproduce – pared with wind-dispersal. i.e. they are relict populations. The expected population turnover Dispersal time for Swedish T. integrifolia popula- It is complicated both to measure and to tions has been estimated by Widén calculate effective dispersal distances. (1987) to 23 years in the Grödby popula- Asteraceae species with pappi are often tion and to 78.6 years in the Benestad believed to be wind-dispersed. Sheldon & population. This would mean that sev- Burrows (1973) have calculated maxi- eral generations of plants would have mum primary dispersal distances for succeeded each other in the Grödby Asteraceae species, among them some population since the 1950s, while many relatively closely related to T. integrifolia of the present Benestad plants could (Senecio squalidus, S. viscosus, & S. have been there since that time. How- vulgaris) to between 1.8 and 2.9 metres ever, there are problems making this at wind speeds of 4.5 metres per second, kind of estimations, since plants that are taking into account among other things not ¥owering can be very inconspicuous, the height of the plants. Dispersal dis- and therefore assumed to be dead. Adult tances are also affected by humidity plants appearing in a patch several years (when the pappus is wet, the seed is after they are believed to have gone ex- likely to land closer to the mother plant), tinct in that spot (Widén pers. obs.) is an and the heat of the surface together with indication that the length of life might surrounding vegetation, both factors earlier have been underestimated. which can affect the vertical component There are several possible explana- of wind direction (Sheldon & Burrows tions for the lack of relationship between 1973). It is possible that the denser veg- population size and genetic variation in etation followed by eutrophication might this study. Similar patterns have been hamper the dispersability of wind-dis- found on species-richness levels – e.g. persed seeds. However, seeds that are Helm, Hanski, & Pärtel (2006) found no traditionally thought to be wind-dis- connection between plant species rich- persed, are often prone to stick to animal ness and present habitat-size in 35 fur (Courvreur et al. 2004, Römermann, alvars in Estonia – however, they did Tackenberg, & Poschlod 2005, De Pablos ¤nd a relationship between present spe- & Peco 2007), and have been shown to be cies richness and landscape connectivity dispersed long distances by sheep 70 years ago. Charlesworth (2009) (Fischer, Poschlod, & Beinlich 1996, makes a review of factors affecting effec- Mouissie, Lengkeek, & Diggelen 2005, tive population size – among them ¥uc- Wessels et al. 2008), cattle (Couvreur et tuations of population size, age struc- al. 2004, Mouissie, Lengkeek, & Digge- ture, and spatial structure (e.g. migra- len 2005), and other large herbivores tion processes). All of these could explain such as donkeys, bison and horses (Rosas the unexpected evenness in variation be- et al. 2008, Couvreur et al. 2004, Couv- tween populations regardless of popula- reur et al. 2008). A study of seed disper- tion sizes – population sizes might have sal by Bison bison in a tallgrass prairie, been very different a hundred years ago. showed that only 17.5% of the seeds If there has not been much recruitment stuck to bison fur had obvious attach- in the smaller T. integrifolia populations ment mechanisms, and that many of the during the last decades, there has been species that are abundant in the fur are little possibility for genetic drift to have usually suspected to be wind-dispersed – any effect on genetic variation, and, in e.g. 32.2% of the seeds were from the addition, the dispersal-mechanism is un- Asteraceae family. In their study of known. For example long-distance dis- diaspore-dispersal by cows, donkeys and persal by animals would give rise to a horses, Couvreur et al. (2004) found that

72 No connection between population size and local genetic variation

24% of the seeds and 42% of the seed- strongly in¥uenced by isolation by dis- lings belonged to the Asteraceae family. tance, and the sampling method is very An over-representation of red-listed important – you can choose either to (Wessels et al. 2008) or declining (Ozinga sample from an area proportional to the et al. 2009) species has been found total area of the population, or you can among epizoochorous plants, and Ozinga choose to sample from an area of exactly et al. (2009) draw the conclusion that the same size in each population. Which ‘colonization de¤cit’ is an important fac- you choose will strongly affect the re- tor in loss of species diversity. sults. According to their simulations If T. integrifolia were indeed wind- (Leblos, Estoup, & Streiff 2006) where dispersed, seeds would be likely to populations of different sizes suffered an spread about a couple of metres each instantaneous loss of individuals and of generation. Since this distance is even habitat area, genetic variation decreased shorter than normal pollination distan- more slowly in populations with strong ces, we would expect extreme genetic isolation by distance than in panmictic structuring within populations, particu- populations, and this effect was stronger larly in fragmented habitats. Small, for a smaller sample area (Leblos, fragmented populations would quickly Estoup, & Streiff 2006). A lack of re- loose genetic variation, and we would cruitment of new individuals might also probably see a difference in genetic vari- render higher-than-expected genetic ation depending on population size, even variation. Van Geert, Van Rossum, & if this relationship might be blurred by Triest (2008) found much lower genetic recent historic changes in population variation in the seedling than in adult size. That we see no relationship at all plants in a Primula vulgaris population, between population size and genetic concluding that habitat degradation variation is a hint either that the now might hamper the establishment of new small populations have until recently individuals. It might be added, that es- been very large and all the same size, or tablishment of new individuals could be that dispersal patterns have been very facilitated if seeds were dispersed over a dynamic. large area, since that would decrease competition from plants of the same spe- Population size and the effect of sample cies, as well as from other plants (re- area viewed in Fenner & Thompson 2005 p. Plants are often more likely to produce 67 ff) – a patch which was ideal for seed- offspring with their nearest neighbours ling establishment when the plants in than with more distant individuals (Tur- the parent generation were established ner, Stephens, & Anderson 1982) while could be very hostile for their offspring. at the same time many plants have bar- riers against producing offspring with Peripheral populations close relatives (Richards 1997, Barrett Populations for preservation are often 2002, Hiscock & Tabah 2003). The pos- selected on a national or state basis. The sibilities of ¤nding mates outside the choices can sometimes seem absurd, population are limited, since plants can’t when you consider that species are for move. Even in a very large population, that reason frequently chosen to be pro- most gene exchange will occur within tected in the periphery of the species dis- very short distances (Turner, Stephens, tribution, where the genetic variation is & Andersson 1982). This adds a dimen- putatively lower than in the centre of the sion to population-genetics not present distribution area, just because a state in most animal populations. border happens to be close to the periph- As pointed out by Leblos, Estoup, & ery of the species distribution (Lesica & Streiff (2006) the effect of reduction in Allendorf 1995). Lammi, Siikamäki, & population size on genetic variance is Mustajärvi (1999) proved small popula-

73 No connection between population size and local genetic variation

tions of Lychnis viscaria to be less gen- Margareta Isaksson for ¤eld assistance in etically variable, but not less ¤t, than Estonia. We thank Maie Jeeser, Toomas larger populations, and drew the con- Kukk, Triin Reitalu, Kaja Riiberg, and Elle clusion that small, isolated populations Valtna for helping us to localise Estonian populations. We particularly thank Andres might harbour important evolutionary Miller and Elle Roosaluste for helping us potential. This is consistent with the access the populations on the Hiiumaa islets. Kadakalaid population in this investi- We thank David Ståhlberg and Stig Isaksson gation, which had least genetic vari- for valuable comments on earlier drafts. ation, but which also differed the most from the other populations. References Abrahams R. 1994. The re-generation of Small-scale genetic variation – causes family farming in Estonia. Sociologia Rura- and consequences lis 34:354–368. The genetic variation in present-day Aguilar R, Quesada M, Ashworth L, Herreri- as-Diego Y, Lobo J. 2008. Genetic conse- populations is the consequence of past quences of habitat fragmentation in plant dispersal and extinction. The relatively populations: susceptible signals in plant high genetic variation in AFLP markers traits and methodological approaches. Mol- suggests that the Swedish T. integrifolia ecular Ecology 17:5177–5188. populations that we see today are relicts Barret SCH. 2002. The evolution of plant of previously large populations, and this sexual diversity. Nature reviews 3:274–284. is con¤rmed by the lack of inbreeding de- Barrett SCH, Kohn JR. 1991. Genetic and pression (Isaksson & Widén paper I) and evolutionary consequences of small population size in plants: implications for con- the high seed-set both in natural popu- servation. In: D.A. Falk and K.E. Holsinger lations (Isaksson & Widén paper I) and (eds.), Genetics and conservation of rare in crossing-experiments (Isaksson paper plants, Oxford University Press, New York, II). If the present Swedish T. integrifolia Oxford (1991), pp. 3–30. populations are indeed relicts of previ- Benvenuti S.2007. Weed seed movement and ously large populations, and if the plants dispersal strategies in the agricultural en- that are still alive are very old, we can- vironment. Weed Biology and Management not expect local genetic variation to be 7:141–157. Charlesworth B. 2009. Effective population lower than in random samples from size and patterns of molecular evolution large and thriving populations. It seems and variation. Nature Reviews Genetics 10: therefore, that the supply of suitable 195–205. mates is suf¤cient for the plants to pro- Coppi A, Mengoni A, Selvi F. 2008. AFLP ¤n- duce high-quality seeds even in the gerprinting of Anchusa (Boraginaceae) in smallest populations in this investiga- the Corso-Sardinian system: Genetic diver- tion (cf. Isaksson & Widén paper I, Isaks- sity, population differentiation and conser- son paper II), at least during years when vation priorities in an insular endemic group threatened with extinction. Biologi- most plants are ¥owering. However, if cal Conservation 141:2000–2011. there have been drastic changes in seed Couvreur M, Verheyen K, Vellend M, Lamoot dispersal dynamics, so that produced I, Cosyns E, Hoffmann M, Hermy M. 2008. seeds are not likely to end up in habitats Epizoochory by large herbivores: merging favourable for seedlings, these seeds will data with models. Basic and Applied Eco- have been produced in vain. logy 9:204–212. Couvreur M, Christiaen B, Verheyen K, Acknowledgements Hermy M. 2004. Large herbivores as mobile links between isolated nature reserves We thank Anna och Svante Murbecks minnes- through adhesive seed dispersal. Applied fond, Lunds botaniska förening, and Elly Ols- Vegetation Science 7:229–236. sons fond for ¤nancial support (to KS). BW De Almeida Vieira F, De Calvalho D. 2008. was supported by FORMAS (grant 21.5/2002- Genetic structure of an insect-pollinated 0115). We thank Mikael Hedrén for introduc- and bird-dispersed tropical tree in vegetation to the lab and valuable advice. We thank tion fragments and corridors: implications

74 No connection between population size and local genetic variation

for conservation. Biodivers. conserv. 17: Kana S, Kull T, Otsus M. 2008. Change in 2305–2321. agriculturally used land and related habitat De Pablos I, Peco B. 2007. Diaspore mor- loss: A case study in eastern Estonia over 50 phology and the potential for attachment to years. Estonian Journal of Ecology 57:119– animal coats in Mediterranean species: an 132. experiment with sheep and cattle coats. Kang M, Jiang M, Huang H. 2005. Genetic Di- Seed Science Research 17:109–114. versity in Fragmented Populations of Dzialuk A, Muchewicz E, Boratyn´ski A, Berchemiella wilsonii var. pubipetiolata Montserrat JM, Boratyn´ska K, Burczyk J. (Rhamnaceae). Annals of Botany 95:1145– 2009. Genetic variation of Pinus uncinata 1151. (Pinaceae) in the Pyrenees determined with Kukk T. 2004. Eesti taimede kukeaabits. cpSSR markers. Plant Syst Evol. 277:197– Tallin. Varak. 205. Kunin WE 1993. Sex and the Single Mustard: Ehrich D. 2006. AFLPdat: a collection of R Population Density and Pollinator Behav- functions for convenient handling of AFLP ior Effects on Seed-Set. Ecology 74:2145– data. Mol. Ecol. Notes 6:603–604. 2160. Ellstrand NC, Elam DR. 1993. Population Lammi A, Siikamäki P, Mustajärvi K. 1999. genetic consequences of small population Genetic Diversity, Population Size, and Fit- size: implications for plant conservation. ness in Central and Peripheral Populations Annual Review of Ecology & Systematics of a Rare Plant Lychnis viscaria. Conserva- 24: 217–242. tion Biology 13:1069–1078. Emanuelsson U, Bergendorff C, Billqvist M, Landergott U, Holderegger R, Kozlowski G, Carlsson B, Lewan N. 2002. Det skånska Schneller, JJ. 2001. Historical bottlenecks kulturlandskapet. 2nd edition. Årsbok för decrease genetic diversity in natural popu- Naturskyddsföreningen i Skåne 2001. Na- lations of Dryopteris cristata. Heredity turskyddsföreningen i Skåne. 87:344–355. Fenner M, Thompson K. 2005. The Ecology of Leblos R, Estoup A, Streiff R. 2006. Genetics Seeds. Cambridge University Press. of recent habitat contraction and reduction Fischer SF, Poschlod P, Beinlich B. 1996. Ex- in population size: does isolation by dis- perimental Studies on the Dispersal of tance matter? Molecular Ecology 15:3601– Plants and Animals on Sheep in Calcareous 3615. Grasslands. The Journal of Applied Ecology Lesica P, Allendorf FW. 1995. When Are 33:1206–1222. Peripheral Populations Valuable for Con- Helm A, Hanski I, Pärtel M. 2006. Slow servation? Conservation Biology 9:753–760. response of plant species richness to habitat Leimu R, Mutikainen P, Koricheva J, Fischer loss and fragmentation. Ecology Letters M. 2006. How general are positive relation- 9:72–77 ships between plant population size, ¤tness Hiscock SJ, Tabah DA. 2003. The different and genetic variation? Journal of Ecology mechanisms of sporophytic self-incompatib- 94:942–952. ility. Philosophical Transactions of the Lodhi MA, Ye G, Weeden NF, Reisch BI. 1994. Royal Society London B 358:1037–1045. A Simple and Ef¤cient Method for DNA Ex- Honnay O, Adriaens D, Coart E, Jacquemyn traction from Grapevine Cultivars and Vitis H, Roldan-Ruiz I. 2007. Genetic diversity Species. Plant Molecular Biology Reporter within and between remnant populations of 12:6–13. the endangered calcareous grassland plant Luijten SH, Dierick A, Gerard J, Oostermeijer Globularia bisnagarica L. Conservation B, Raijmann LEL, Den Nijs, HCM. 2000. Genetics 8:293–303. Population size, genetic variation, and re- Honnay O, Jacquemyn H. 2007. Susceptib- productive success in a rapidly declining, ility of common and Rare Plant Species to self-incompatible perennial (Arnica monta- the Genetic Consequences of Habitat Frag- na) in The Netherlands. Conservation Bio- mentation. Conservation Biology 21:823– logy 14:1776–1787. 831. Mhemmed G, Kamel H, Chedly A. 2008. Does Ingelög T, Andersson R, Thernborg M. (eds.) habitat fragmentation reduce genetic diver- 1993. Red Data Book of the Baltic Region. sity and subpopulation connectivity? Eco- Part 1. Lists of threatened vascular plants graphy 31:751–756. and vertebrates. Swedish Threatened Spe- Mouissie AM, Lengkeek W, Van Diggelen R. cies Unit, Uppsala, in cooperation with the 2005. Estimating adhesive seed-dispersal Institute of Biologya, Riga. distances: ¤eld experiments and correlated

75 No connection between population size and local genetic variation

random walks. Functional Ecology 19:478– of Chionanthus retusus (Oleaceae) of 486. Tsushima Island and the Tôno region, Ouborg NJ, Piquot Y, Van Groenendael JM. Japan. Genes Genet. Syst. 73:29–37. 1999. Population genetics, molecular mark- Stoate C, Boatman ND, Borralho RJ, Rio ers and the study of dispersal in plants. Carvalho C, De Snoo GR, Eden P. 2001. Eco- Journal of Ecology 87:551–568. logical impacts of arable intensi¤cation in Ozinga W, Römermann C, Bekker RM, Europe. Journal of Environmental Manage- Prinzing A, Tamis WLM, Schaminée JHJ, ment 63:337–365. Henneskens SM, Thompson K, Poschlod P, Tomimatsu H, Ohara M. 2006. Evaluating the Kleyer M, Bakker JP, Van Groenendael JM. consequences of habitat fragmentation: a 2009. Dispersal failure contributes to plant case study in the common forest herb Tril- losses in NW Europe. Ecology Letters lium camschatcense. Population Ecology 12:66–74. 48:189–198. Peterson U, Aunap R. 1998. Changes in agri- Turner ME, Stephens JC, Anderson WW. cultural land use in Estonia in the 1990s de- 1982. Homozygosity and Patch Structure in tected with multitemporal Landsat MSSS Plant Populations as a Result of Nearest- imagery. Landscape and Urban Planning Neighbor Pollination. PNAS 79:203–207. 41:193–201 Van Geert A, Van Rossum A, Triest L. 2008. R Development Core Team 2008. R Version Genetic diversity in adult and seedling 2.7.0. The R Foundation for Statistical populations of Primula vulgaris in a frag- Computing. ISBN 3-900051-07-0. Available mented agricultural landscape. Conserva- on line at http://www.r-project.org/ tion Genetics 9:845–853. Richards AJ. 1997. Plant Breeding Systems. Vos P, Hogers R, Bleeker M, Reijans M, Van Second edition. Chapman & Hall. De Lee T, Hornes M, Freijters A, Pot J, Rohlf FJ. 2005. NTSYS-pc, numerical tax- Peleman J, Kuiper M, Zabeau M. 1995. onomy and multivariate analysis system, AFLP, a new technique for DNA ¤nger- version 2.2 Exeter software, Setauket, New printing. Nucleic Acids Research 23:4407– York, USA. 4414. Römermann C, Tackenberg, O, Poschlod P. Wessels S, Eichberg C, Storm C, Schwabe A, 2005. How to predict attachment potential 2008. Do plant-community-based grazing of seeds to sheep and cattle coat from regimes lead to epizoochorous dispersal of simple morphological seed traits. Oikos high proportions of target species? Flora 110:219–230. 203:304–326. Rosas CA, Engle DM, Shaw JH, Palmer MW. Widén B. 1987. Population biology of Senecio 2008. Seed dispersal by Bison bison in a integrifolius (Compositae) a rare plant in tallgrass prairie. Journal of Vegetation Sweden. Nord. J. Bot. 7:687–704. Science 19:769–778. Widén B. 1993. Demographic and genetic ef- Schneider S, Roessli D, Excof¤er L. 2000. fects on reproduction as related to popula- Arlequin version 2.000: a software for popu- tion size in a rare, perennial herb, Senecio lation genetic data analysis. Genetics and integrifolius (Asteraceae). Biological Jour- Biometry Laboratory, University of Gene- nal of the Linnean Society 50:179–195. va, Geneva, Switzerland. http://cmpg. Widén B, Wetterin M. 1999. Åtgärdsprogram unibe.ch/software/arlequin3. för bevarande av fältnockan (Tephroseris in- Sheldon JC, Burrows FM. 1973. The Dis- tegrifolia). Naturvårdsverket. Stockholm. persal Effectiveness of the Achene – Pappus Wilcock C, Neiland R. 2002. Pollination fail- Units of Selected Compositae in Steady ure in plants: why it happens and when it Winds with Convection. New Phytologist matters. TRENDS in Plant Science 7:270– 72:665–675. 277. Smith UK. 1979. Biological ¥ora of the British Yeh FC, Yang RC, Boyle T. 1999. POPGENE. Isles, Senecio integrifolius (L.) Clairv. Jour- Microsoft windowsbased freeware for popu- nal of Ecology 67:1109–1124. lation genetic analysis. Release 1.31. Uni- Soejima A, Maik M, Ueda K. 1998. Genetic versity of Alberta, Edmonton and Centre for variation in relic and isolated populations International Forestry Research.

76 IV Negative association between population size and quantitative genetic variance

78 Negative association between population size and quantitative genetic variance

Negative association between population size and quantitative genetic variance in Tephroseris integrifolia (Asteraceae) – evidence for a recent bottleneck?

Kerstin Isaksson and Björn Widén, Department of Ecology, Section Plant Ecology and Systematics, Lund University, Sweden.

Abstract The Field Fleawort, Tephroseris integrifolia (L.) Holub, is a plant species con¤ned to nutrient- poor, calcareous grasslands, a habitat-type which has declined considerably in many European countries during the past 150 years. As a result, T. integrifolia has become very rare in some of these countries. We investigated the broad sense heritability for T. integrifolia plants from thirteen North European populations, from Britain, Denmark, Sweden, and Estonia. Surprisingly, heritability was for most characters the largest in the Swedish populations, which are the smallest and the smallest for the Estonian populations, which are the largest. It is probable that the smaller populations are relicts of large populations, which could explain why not much variation has been lost. Increased quantitative variation could, however, be the consequence of homozygosity in rare recessive alleles, something which has been observed during the ¤rst generations of a bottleneck. It seems, therefore, that not many generations of T. integrifolia have passed since the populations decreased in size, something which is consistent with other studies of the same species.

Keywords: Quantitative genetics, genetic variation, small populations, fragmentation, extinction, relict populations, threatened species.

Introduction 1940, when the use of tractors and arti- A large number of the plant species in ¤cial fertilisers became more common, Europe are native to nutrient-poor grass- changing most of the remaining grasslands (Pärtel et al. 2005). This is a habi- lands into a completely new habitat type tat-type which has been dependent on (Ratcliffe 1984, Bernes 1994). Gradually, human management for thousands of agriculture has become more ef¤cient years (Richards 1990). During the past and also more static, through the cessa- centuries there has been considerable tion of the rotation of crops and the more change in the European cultural land- rigid division of land into ¤elds and pas- scape. Approximately 30% of the total tures, and as a consequence the area grassland-cover in (non USSR) Europe covered with nutrient-poor grasslands disappeared between 1700 and 1980, has decreased in Europe (Emanuelsson most of which took place before 1850 2009). Therefore, populations of plants (Richards 1990). However, the more se- native to these habitats have become vere changes have taken place after fragmented and many are threatened

79 Negative association between population size and quantitative genetic variance

with extinction (Hodgson et al. 2005, possibly detrimental, alleles increases Poschlod et al. 2005, Schleuning et al. with increased homozygosity (Reed & 2009). Several studies have demon- Frankham 2003). strated that the effect of fragmentation Reduced variation in quantitative on species richness or population size characters as a consequence of decreased might be delayed, so that we don’t yet see population size are well documented in the full effect of changes that took place experimentally inbred populations, at 50 or 100 (Cousins, Ohlson, & Eriksson least for characters not strongly related 2007), or even 150 (Tasser et al. 2007), or to ¤tness – which is demonstrated in a 200 years ago (Gustavsson, Lennartson, meta-study of 22 investigations by Van & Emanuelsson 2007). The processes of Buskirk & Willi (2006) – but are weakly change have, due to political and econ- supported by studies of natural popula- omical differences, varied between coun- tions, according to a review by Willi, Van tries, particularly, many processes have Buskirk, & Hoffmann (2006). Variation started later in the eastern and central in quantitative markers is sometimes parts of Europe as a consequence of a found to be negatively correlated with more rigid feudal system (Emanuelsson population size in natural populations 2009). (e.g. Waldmann 2001) and has been noted in the ¤rst generations of in- Small populations breeding in experimental populations Stochastic events – demographic or envi- (e.g. Van Heerwarden et al. 2008). Unex- ronmental catastrophes – may lead to pectedly high levels of genetic variation the extinction of small populations (Tu- have been found in several studies of relli 1977, Vindens, Engen, & Sæter small populations and could be ex- 2008). Small populations loose genetic plained by the populations being relicts variance due to genetic drift (Young, of larger populations (e.g. Young, Brown, Brown, & Zich 1999, Van Rossum, De & Zich 1999, Luijten et al. 2000). In the Sousa & Triest 2004). Genetic depletion meta-study by Van Buskirk & Willi can be observed in molecular, putatively (2006), it is found that variation in ¤t- neutral, markers, e.g. as the loss of het- ness-related morphological characters erozygosity, which will be 1/[2Ne] per often increases during the ¤rst gener- generation (Wright 1931), where Ne is ations of inbreeding – a consequence, the effective population size – thus, the they conclude, of dominance and epi- smaller the population, the larger the stasis effects on life history traits. loss of genetic variance. Genetic deple- While organisms chosen for bottleneck tion is also re¥ected in the phenotypic experiments usually have a relatively variance and leads to a decrease in ad- short generation time (e.g. Bryant & ditive genetic variance, VA=VP/[VD+VE], Meffert 1988, Saccheri, Nichols, & where VP is the phenotypic variance in Brake¤eld 2006), generation times in na- the population, VD the variance of domi- tural populations vary and for long-lived nance deviations and VE the variance species, a hundred years after habitat due to environmental effects (Futuyma fragmentation, we might only see the 1998). If there is no epistasis, dominance effects of a few generations of in- or selection, the loss of additive genetic breeding, which is when we might expect variance will also be 1/[2Ne] per gener- an increase in variation in some quanti- ation. A loss of phenotypic variance leads tative characters (Van Buskirk & Willi to a reduced capacity to tackle future en- 2006). The period of increasing variation vironmental changes as a consequence of at the beginning of a bottleneck might be decreased evolvability (Houle 1989, Reed prolonged if the mechanisms behind the & Frankham 2003). Genetic depletion decline of the populations also hamper also increases the risk of inbreeding de- successful regeneration, in which case pression; as the expression of recessive, extant populations are relicts of larger,

80 Negative association between population size and quantitative genetic variance

historical populations (Tilman et al. tries – Britain, Denmark, Sweden, and 1994, Bachmann & Hensen 2007, García Estonia. The decrease in the Swedish T. 2008). integrifolia populations is closely associ- The Field ¥eawort, Tephroseris integri- ated with the disappearance of pastures folia, is one of the plant species which is and the rationalisation of agriculture con¤ned to nutrient-poor grasslands and (Isaksson & Widén paper I). While agri- its decline in Sweden has been studied in cultural history in Britain, Denmark, detail (Widén 1986, 1987, 1991, 1993, and Sweden have followed more or less Widén & Andersson 1993, Widen & the same processes as the rest of West- Wetterin 1999, see also Isaksson & ern Europe, Estonian agrarian history is Widén paper I, Isaksson paper II, Isaks- more complicated, since Estonia has son, Nordström, & Widén paper III). De- been under the rule of so many different spite a serious decline in population regimes (Emanuelsson 2009). According number and size (Isaksson & Widén pa- to Pärtel et al. (1999), the destruction of per I), the small Swedish populations semi-natural grass-lands has started to- don’t suffer from serious inbreeding de- wards the end of the ¤fties – however, pression (Isaksson & Widén paper I), Estonian landscape has gone through when cross-pollinated with plants from several periods of change during the 20th the same patch, they usually set seed, century. By the time of world war one, though they are self-incompatible, which most agricultural land in Estonia was indicates that they have not lost many S- owned by German-Baltic nobility. In the alleles through drift (Isaksson paper II) 1920s, agricultural land was sold to and the Swedish populations seem to farmers and towards the end of the harbour about as much local genetic 1940s, under Soviet regime, it went variation in molecular markers as the through collectivisation, which led to a larger Estonian populations (Isaksson, lot of smaller farms being abandoned Nordström, & Widén paper III). (Emanuelsson 2009). However, in While the study of variation in puta- Estonia, the picture is further compli- tively neutral genetic markers is a good cated by the custom of small private tool to ¤nd loss of genetic variation in plots where small-scale farming was small populations, it is also important to done parallel to large-scale collective investigate the morphological response farming during the Soviet period (Abra- to genetic variation, i.e. additive genetic hams 1994). In the 1970s and 1980s, variance, or heritability. In an earlier about 15% of Estonia was turned into study of two Swedish populations of T. nature reserves (Emanuelsson 2009). integrifolia, a smaller population (Bene- Additional semi-natural grasslands in stad) seemed to harbour more variation Estonia are now threatened as a conse- than a larger population (Grödby) (Wi- quence of many farms being abandoned dén & Andersson 1993). However, all in the 1990s, after the Soviet era Swedish populations are relatively small (Emanuelsson 2009). e.g. compared with the Estonian popula- We use T. integrifolia as a model or- tions (Isaksson pers. obs.) and also have ganism to answer the following ques- a relatively similar history. We wanted to tions: make a broader study of populations • What is the relationship between from a greater geographical range, with population size and variation in a more varied history and with more quantitative characters? varying numbers of plants. • Is there a difference between North In this study we investigate the rela- European regions in local or re- tionship between population size and gional genetic variation? variation in morphological characters in thirteen North European populations of T. integrifolia, from four different coun-

81 Negative association between population size and quantitative genetic variance

Materials and methods (Den danske rødliste 2004) or Estonia. Study species However, in Red Data Book of the Baltic Tephroseris integrifolia L. Holub (Aster- Region (Ingelög et al. 1993), T. integri- aceae – earlier Senecio integrifolius) is a folia is listed as rare in Denmark, while perennial, self-incompatible herb which in Estonia it is listed as present. grows in nutrient-poor, calcareous grasslands. It occurs throughout Eurasia, Populations from Britain to Japan. In several Euro- For information on size, structure, and pean countries it has declined consider- cover-area of the populations, see ably during the past decades (e.g. Widén Table1! On the three British sites, the 1987, Krach & Krach 1991, and Preston plants grew on slopes related to pre-his- et al. 2002 compared with Smith 1979). toric constructions. The smallest popula- Each stem has 1-4 ¥ower heads with 30- tion (Cissbury Ring) was infested by rab- 150 ¥orets in each head. The fruit is a bits. All three sites were grazed. Of the one-seeded achene with a pappus. In the three Danish sites, two (Sårup and wild, a plant rarely ¥owers two years in a Skørping) were grazed. In both cases the row and the number of ¥owering plants plants grew on steep slopes, which would in one population varies considerably be- be avoided by cattle as long as there is tween years (Isaksson & Widén paper I). more easily accessible grass. At one site The Swedish populations have been (Hamborg) there was no grazing – how- carefully monitored and investigated ever, the slope was so steep, that vegeta- since the 1980s (e.g. Widén 1986, 1987, tion was probably kept open through 1993, Andersson & Widén 1994) and du- continuous erosion (Isaksson, pers. obs.). ring this period they have declined dras- Of the four Estonian populations, none tically in spite of extensive conservation grew on a slope. In the westernmost measures. Observed pollinators are population, at Vohilaid (an island to the hover ¥ies and other ¥ies (Widén pers. east of the island Hiumaa) the plants obs.), which usually don’t ¥y long dis- grew among lime stone gravel on the tances. This explains why seed-set is ex- shore. The plants in the population at ceedingly density dependent, which has Pakri neem grew together with high been demonstrated in a common garden grass close to a steep slope towards the experiment (Widén 1993), and why sea. We saw no signs of grazing. At plants growing more than ten metres Allika, all the plants grew along the away from others of the same species edges of a gravel-road with over-grown seldom set any seeds (Isaksson & Widén pastures on both sides – indicating that paper I). Widén (1987) calculated the ex- the extant plants are a relict of a much pected half-life of plants in four Swedish larger population. The Muuksi linna- T. integrifolia populations to between 7.2 mägi population, which was by far the and 39.3 years. One of these populations largest, grew within a small area which (Tosteberga) is included in this study, had been grazed the previous year with an expected half-life of 8.7 years for (which we concluded from the year-old individual plants. cowpats). In 1983 the Tosteberga population was the largest Swedish population, Regions with about 8000 ¥owering stems, and We sampled seeds from thirteen North new seedlings were established. Ten European populations in Britain, Den- years later the size of the population had mark, Sweden, and Estonia. In Britain decreased dramatically (Isaksson & Wi- (Chef¤ngs & Farrell 2005) and in Swe- dén paper I). During the past ten years den (Gärdenfors 2005) this species is re- the maximum number of ¥owering stems garded as endangered (EN according to in Tosteberga has been 200, in Edenryd IUCN criteria), while it is not present on 600, and in Råby 800 (Isaksson & Widén the national red data lists for Denmark paper I).

82 Negative association between population size and quantitative genetic variance

Table 1. Description of T. integrifolia populations. Populations in Denmark, Britain, and Swe- den were sampled in 2006, and Estonian populations in 2005. Covered area and population structure were estimated from the distribution of flowering plants in the area during the sampling year.

Region Population Coordinates Covered Nr of Population Density Habitat N. of field area flowering structure (stems/m 2) appearance mothers stems collected from 57°06’48N Denmark Hamborg 200 m 2 40 Patchy 0.2 Slope 11 8°40’21E 56°50’28N Skørping 30 m 2 17 Patchy 0.6 Slope 8 9°54’52E 57°05’8N Mostly Sårup 1500 m 2 200 0.1 Slope 9 8°38’59E continuous

50°51’45N Britain Cissbury Ring 140 m 2 24 Patchy 0.2 Slope 10 0°22’58W 50°51’42N Continuous, Mt Caburn 1800 m 2 1800 1.0 Slope 10 0°22’58W oval 50°58’14N Continuous, Martin Down 400 m 2 142 0.4 Slope 10 0°03’3E linear

59°23’31N Gravel Estonia Allika 200 m 2 30 Linear 5.5 12 24°23’24E roadside Muuksi 59°30’41N Continuous, 0,5 km 2 5 mill. 20 Pasture, flat 12 linnamägi 25°31’32E circular 59°23’23N Mostly Grassland, Pakri neem 16000 m 2 230 000 14.4 12 24°02’37E continuous flat 58°55’29N Chalk gravel Vohilaid 2000 m 2 3500 Continuous 1.8 10 23°01’41E shore

Continuous + 56°04’87N Flat, stony Sweden Edenryd 800 m 2 228 some isolated 0.3 10 14°51’95E pasture patches

56°09’69N Råby 700 m 2 152 Patchy 0.2 Flat pasture 10 14°51’79E 56°01’65N Flat, stony Tosteberga 1000 m 2 164 ±Continuous 0.2 10 14°46’09E pasture

Sampling and ¤eld-measures divided between the blocks. In December We sampled between eight and twelve all plants were randomized and classi- plants from each population (Table 1). To ¤ed according to size, 0 through 5, and make the populations comparable, we we measured the longest leaf of each sampled from equal-sized plots (ca. 10 plant. The smallest plants, category 0, m2). From each plant we took one ¥ower were kept in the warm green-house head with mature seeds. Height of the throughout the winter, while the rest tallest stem for each plant was measured were kept in a cold green-house from (except for Mt Caburn) and number of December, when the plants were ran- ¥owering stems per plant and ¥ower domized within blocks, till February, heads per stem were counted. when the plants were again randomized within blocks and transferred back to the Cultivation of offspring warm green-house. All seeds were sown during two days in August 2006. From each of the ten ¤eld- Measures mother plants from each population, we Characters chosen for measure (Tables 3 sowed 30 seeds, or all seeds, when less and 4) had been found to give signi¤cant than 30 were available and transplanted narrow sense heritabilities in an earlier 20 seedlings, or all seedlings, when less study of the same species (Widén & An- than 20 were available, into new pots. dersson 1993) – except the characters The seedlings were transplanted again »number of undeveloped ¥ower heads» in September. All plants were divided and corymb length, which were not in- into (and randomized within) blocks, cluded in Widén and Andersson (1993). with members of each family evenly The ¥owering date of the ¤rst ¥ower

83 Negative association between population size and quantitative genetic variance

Fig. 1. Heights at flowering of T. Integrifolia in cultivation for different populations. Colour indicates homogeneous sub-set at p<0.05; white=subset 1, gray=subset 2, black = subset 3. head of each plant was recorded and this and were measured, on average 69 from was also the date when all measures each population (Tab. 2). were taken (except ¥ower- and seed-characters and height at seed-maturity). Heritabilities and QSTs One ¥ower head from each plant was Nested variance component analyses harvested and preserved in 40% ethanol were done for one character and one in a cold store-room. The plants were country at the time. For ¥owering time randomly cross-pollinated within popu- and height, populations were also ana- lation in order to make the plants pro- lysed separately. Variance components of duce seeds. The date of seed-maturity populations V(p), variance components was recorded and from each plant one of ¤eld mothers, nested within popula- head with seeds was harvested. Flower- tions V(m(p)), and variance components and seed-measures were taken with an of errors V(e) were given by the computer ocular ruler under a dissecting micro- analysis. The genetic variance was quan- scope and with a digital slide-calliper. ti¤ed as the intra-class coef¤cient, calcu- Characters were grouped into ¤ve cate- lated as (Falconer & Mackay 1996): gories: height characters, quantity char- V(m(p))/(V(m(p) + V(e)) acters, ¥ower characters, seed charac- and referred to as the broad-sense ters, and phenology characters (¥ow- heritability, H2. We do not know whether ering time, seeding time, and length of plants from one family are full-sibs or ¥owering period). In total 1994 plants half-sibs which would imply multiplying were cultivated, of which 896 ¥owered the intra-class coef¤cient with 2 or 4 to

84 Negative association between population size and quantitative genetic variance

Fig. 2. Flowering period of the different populations, calculated from the first flowering date and the first seeding date of individual plants.

Table 2. Average first flowering date. Posthoc: Tukey HSD (homogeneous sub-sets with significance-values).

Country Population N field mothers N flowering Average first Subset (p) Subset Subset (p) with flowering offspring flowering (p) offspring Estonia Allika 11 71 April 5 1 (0,837) Estonia Pakri neem 11 58 April 7 1 Sweden Tosteberga 10 104 April 13 2 Estonia Vohilaid 10 47 April 14 2 Estonia Muuksi 10 63 April 14 2 (0,053) 3 Sweden Edenryd 10 86 April 17 2 3 (0,054) 4(0,597) Sweden Råby 9 93 April 18 2 3 4 Denmark Sårup 9 76 April 19 3 (0,054) 4(0,597) 5(0,058) Britain Mt Caburn 10 49 April 19 4(0,597) 5(0,058) Denmark Hamborg 9 75 April 20 4 5 Britain Cissbury Ring 10 66 April 23 5(0,058) 6 (0,411) Britain Martin Down 10 60 April 27 6 (0,411) 7 (0,989) Denmark Skørping 8 48 April 29 7 (0,989)

get narrow-sense or broad sense heritab- QST=V(p)/V(p) +2 × (V(m(p) )+ V(e)) ility, respectively. We use the term broad- One of the Danish populations (Ham- sense heritability, to imply that we can- borg) consisted of very small plants, not decide how much of the variation is which made the QST -values for Denmark due to additive variance or dominance very high, therefore, we also calculated variance, respectively. QST for the Swedish and Danish popula- QST for each country was calculated as tions together, excluding the Hamborg (cf. Spitze 1993): population.

85 Negative association between population size and quantitative genetic variance

Table 3. Heritabilities. Significance values: * for p<0.05, ** for p<0.01, *** for p<0.001 reflect significance-values of corresponding ANOVAs. Average values bold when at least one individual character within the character category is significant. H2-values Denmark Britain Estonia Sweden Height characters Height at flowering 0.188 *** 0.000 0.027 0.104 * Height after flowering 0.093 * 0.018 0.000 0.086 * Length of first head stalk 0.076 * 0.187 * 0.030 0.126 * Length of corymb 0.105 * 0.313 * 0.194 * 0.091 * Average 0.116 0.130 0.063 0.102

Quantity characters Total number of heads 0.093 * 0.000 0.050 0.113 * Tot undeveloped heads 0.000 0.000 0.000 0.000 Number of ray florets 0.074 * 0.083 0.000 0.307 * Number of disc florets 0.054 0.123 * 0.024 0.222 * Average 0.055 0.051 0.018 0.160

Flower characters Ray floret width 0.138 * 0.269 * 0.147 * 0.199 * Ligule length 0.099 * 0.091 * 0.121 * 0.346 * Corolla length 0.122 * 0.096 * 0.000 0.276 * Style length 0.144 * 0.001 0.038 0.087 * Average 0.126 0.114 0.077 0.227

Seed characters Seed length 0.152 * 0.222 * 0.059 * 0.191 * Seed width 0.242 * 0.193 * 0.101 * 0.091 * Pappus length 0.000 0.177 0.149 * 0.120 * Average 0.131 0.197 0.103 0.134

Phenology characters First flowering day 0.027 * 0.104 * 0.000 0.028 First seeding day 0.011 * 0.069 * 0.033 * 0.080 Length of flowering period 0.000 0.046 0.000 0.056 Average 0.013 0.073 0.011 0.055 Number characters with 14 10 7 14 significant values

All analyses were performed in SPSS Two of the Estonian populations, 16.0 (SPSS Inc., Chicago, IL). Allika and Pakri neem, ¥owered signi¤- cantly earlier than the rest. The Est- Results onian and the Swedish populations all P-values are given in tables and ¤gures. ¥owered earlier than the Danish and the British populations, but the ¥owering Differences between populations was to a large extent overlapping The plants from the Estonian popula- between all of the populations (Fig.2, tions were taller than the rest and the Table 2). plants from one of the Danish popula- Generally, the ¥owering period seemed tions (Hamborg) were the shortest correlated with area of origin, as plants (Fig.1). originating from geographically close

86 Negative association between population size and quantitative genetic variance

Table 4. QST-values. Significance values: * for p<0.05, ** for p<0.01, *** for p<0.001 reflect significance-values of corresponding ANOVAs. Average values bold when at least one individual character within the character category is significant.

Qst-values Denmark Britain Estonia Sweden 5 North. pop. Height characters Height at flowering 0.354 ** 0 0 0 0.039 ** Height after flowering 0.435 * 0.006 0.011 0 0.027 ** Length of first head stalk 0.378 * 0 0.009 0.031 * 0.100 *** Length of corymb 0.141 * 0 0 0.006 0.038 ** Average 0.327 0.002 0.005 0.009 0.051

Quantity characters Total number of heads 0.081 * 0.047 * 0.02 0 0.00075 Tot undeveloped heads 0.121 * 0 0.034 0 0 Number of ray florets 0.202 * 0.026 0.017 0.027 * 0.087 *** Number of disc florets 0.092 * 0 0 0.009 0.073 *** Average 0.124 0.018 0.018 0.009 0.040

Flower characters Ray floret width 0.08 0 0.083 0 0 Ligule length 0.139 * 0 0.068 * 0 0 Corolla length 0.056 * 0.129 * 0 0 0.029 Style length 0.029 0.089 * 0 0 0.037 * Average 0.076 0.054 0.038 0 0.0165

Seed characters Seed length 0.038 0.141 * 0.1 * 0.015 0.013 Seed width 0.019 0.052 * 0 0 0.00011 Pappus length 0.101 * 0.158 * 0.005 * 0.072 * 0.077 Average 0.053 0.117 0.035 0.029 0.030

Phenology characters First flowering day 0.06 0.117 * 0.106 0 0.073 *** First seeding day 0.123 0.109 * 0.06 0 0.128 *** Length of flowering period 0.006 0.009 0.012 0 0.0014 Average 0.063 0.078 0.06 0 0.0675 Number of characters with 11 8 3 3 9 significant values

populations also ¥owered at the same in at least one of the four regions. Herit- time. The plants from the Danish popu- ability was signi¤cant for 14 characters lations Hamborg and Sårup, which were in the Danish and Swedish regions, for morphologically very different, probably 10 characters in England and for 7 char- due to one or a few genes causing the acters in Estonia (Table 3). Hamborg plants to be very small, were For most morphological characters, much closer in ¥owering period than values of heritability were the largest for either of them were with the plants from the Swedish populations and the smal- Skørping. All three Swedish populations, lest for the Estonian populations (Fig.3). which in the ¤eld grow within eleven For ¥owering time, only four populations kilometres from each other, seemed syn- (Sårup, Cissbury Ring, Martin Down, nchronized in their ¥owering (Fig.2, and Tosteberga), showed any signi¤cant Table2). heritability (¤gures not shown).

Heritability Differences in QST between regions Of the 18 characters investigated for QST values (Table 4 and Fig.4) were heritability, 16 showed signi¤cant values higher for the Danish populations than

87 Negative association between population size and quantitative genetic variance

Fig. 3. Average heritability for all measured characters (calculated from the means of the different categories given in Table 2). for the rest – particularly for characters Widén paper I), while the Estonian popu- related to height and characters which lations are still large and seem to be might be affected by height. When Ham- thriving (Isaksson pers. obs.). However, borg was removed from the analysis, and our results are consistent with an earlier the Swedish and Danish populations study of T. integrifolia; Widén & Anders- were analysed together, then QST for the son (1993) found – in a paternal half-sib Nordic region was about the same as for analysis – a small population to have the other regions. Thus, it is likely that both higher heritabilities and signi¤cant the Swedish and the Danish populations additive genetic variance for a larger are relatively closely related, although number of traits than a large population they are separated by several straits of (two Swedish populations not used in the water. For Estonia, though the QST present study). They investigated 35 values were generally high compared characters related to morphology and with the other regions, only three values life-history, and 19 of them displayed sig- were signi¤cant. ni¤cant heritability in the small population and 14 in the large population. The Discussion magnitude of the obtained heritability Our study demonstrates higher heritab- values were comparable to values esti- ility values and signi¤cant heritability mated for common and wide-spread for more traits in the Swedish popula- plants, suggesting that the two investi- tions than in the Estonian populations – gated populations of T. integrifolia had an unexpected result, since the Swedish not suffered from genetic erosion (Widén populations have been small and & Andersson 1993). threatened for a long time (Isaksson & The estimated population sizes in the

88 Negative association between population size and quantitative genetic variance

Fig. 4. Average QST for all measured characters (calculated from the means of the different categories given in Table 2). The region 5 Nordic populations comprise the Swedish and Danish populations except the morphologically different Hamborg population. present paper are based on the number genetic variation is a persistent seed- of ¥owering stems during the sampling bank (Honnay et al. 2008). If habitat de- year. However, individual plants rarely struction hampers regeneration, while ¥ower every year (Widén 1987) and until adult individuals survive, the time from a few years ago, the number of ¥owering habitat destruction to population extinc- stems were much higher in the Swedish tion depends on how long-lived individ- populations than they are now (Isaksson uals are – if they can grow very old we & Widén paper I). Consequently, the might be given the false impression that number of ¥owering stems during a par- the population is still viable, while it is ticular year is not an exact measure of in fact a relict population condemned to population size, but would give a rough extinction, unless the habitat is restored. indication. Obligately outcrossing populations usu- Unexpectedly high genetic variation in ally maintain genetic variance for longer small populations, albeit lower than in – unexpectedly high genetic variation is large populations is often associated often associated with self-incompatib- with recent decreases in population size ility (Colas et al. 1997, Young, Brown & – drift has not yet taken its toll on Zich 1999, Luijten et al. 2007, Campbell genetic variance – the population is then et al. 2007, Honnay et al. 2007, Mathia- said to be a relict (e.g. Young, Brown & sen et al. 2007) as well as longevity Zich 1999, Luijten et al. 2000). The time (Soejima et al. 1998, Dzialuk et al. 2009, from habitat destruction to population De Almeida Vieira & De Carvalho 2008). extinction varies between species. An- A short-term effect of a decrease in other explanation of unexpectedly high population size could even lead to an in-

89 Negative association between population size and quantitative genetic variance

crease in heritability as a result of in- References creased homozygosity in previously rare Abrahams R. 1994. The re-generation of fam- recessive alleles (Willis & Orr 1993), ily farming in Estonia. Sociologia Rurali something which, according to a meta- 34:354–368. study by Van Buskirk & Willi (2006) has Bachmann U, Hensen I. 2007. Is declining Campanula glomerata threatened by gen- been con¤rmed in several empirical etic factors? Plant Species Biology 22:1–10. studies. We have not found any studies Bernes, C.1994. Biologisk mångfald i Sve- where genetic variation in molecular rige: en landstudie. Monitor 14, Statens na- markers has been reported to be higher turvårdsverk, Solna. in small populations than in large popu- Bryant EH, Meffert LM. 1988. Effect of an Ex- lations. perimental Bottleneck on Morphological In- Our results indicate that the small tegration in the House¥y. Evolution Swedish T. integrifolia populations are 42:698–707. Campbell LG, Husband BC.2007. Small going through a bottleneck, but that they populations are mate-poor but pollinator- have not been small for many gener- rich in a rare, self-incompatible plant, Hy- ations. According to papers I-III in this menoxys herbacea (Asteraceae). New Phyto- thesis, Swedish T. integrifolia popula- logist 174:915–925. tions show no obvious signs of loss of gen- Chef¤ngs CM, Farrell L. (eds), Dines TD, etic variation – something which is ex- Jones RA, Leach SJ, McKean DR, Pearman pected several generations after frag- DA, Preston CD, Rumsey FJ, Taylor I. mentation – however, the populations 2005. The Vascular Plant Red Data List for Great Britain. Species Status 7: 1–116. have decreased in size. Unlike the other Joint Nature Conservation Committee, three studies in this thesis, this study in- Peterborough. dicates population decrease has had an Colas B, Olivieri K, Riba M. 1997. Centaurea effect on genetic variation in Swedish T. corymbosa, a cliff-dwelling species tottering integrifolia populations, although con- on the brink of extinction: A demographic sidering how quickly the Swedish popu- and genetic study. Proc. Natl. Acad. Sci. lations have decreased in size during the 94:3471–3476. 20th century, it is likely that they will dis- Cousins SAO, Ohlson H, Eriksson O. 2007. Effects of historical and present fragmenta- appear before these changes will have tion on plant species diversity in semi-na- any detrimental effect on the plants. tural grasslands in Swedish rural land- Acknowledgements scapes. Landscape Ecology 22:723–730. Financial support to BW was provided by De Almeida Vieira F, De Calvalho D. 2008. FORMAS (grant no 21.5/2002-0115). We Genetic structure of an insect-pollinated thank Anna och Svante Murbecks minnes- and bird-dispersed tropical tree in vegeta- fond, Lunds botaniska förening, and Elly Ols- tion fragments and corridors: implications sons fond for ¤nancial support to KI. We for conservation. Biodivers. conserv. thank Stefan Andersson for advice concerning 17:2305–2321. experiment design and data analysis. We Den danske rødliste 2004. Den danske rødlis- thank Maie Jeeser, Toomas Kukk, Triin te. Fagdatacenter for Biodiversitet og Ter- Reitalu, Kaja Riiberg, and Elle Valtna for restrisk Natur (B-FDC). – Danmarks helping us to localise Estonian populations. Miljøundersøgelser. Found at: http:// We particularly thank Andres Miller and Elle redlist.dmu.dk. Roosaluste for helping us access the popula- Dzialuk A, Muchewicz E, Boratyn´ski A, tions on the Hiiumaa islets. We thank Alan Montserrat JM, Boratyn´ska K, Burczyk J. Knapp, Tony Mundel, David Pearman, and 2009. Genetic variation of Pinus uncinata Sharon Pilkington for helping us to localise (Pinaceae) in the Pyrenees determined with British populations. We thank Hans Henrik cpSSR markers. Plant Syst Evol. 277:197– Bruun for helping us localise Danish popula- 205. tions. We thank Margareta Isaksson for ¤eld Emanuelsson U. 2009. Europeiska kultur- assistance in Estonia and Britain and David landskap. Hur människan format Europas Ståhlberg for ¤eld assistance in Denmark. We natur. Forskningsrådet Formas. thank Stefan Andersson and David Ståhlberg Falconer DS, Mackay FC.1996. Introduction for valuable comments on earlier drafts. to Quantitative Genetics. Longman group Ltd.

90 Negative association between population size and quantitative genetic variance

Futuyma DJ. 1998. Evolutionary Biology. diversity in temperate European grass- Third Edition. Sinauer Associates, Inc. Pub- lands: origin and conservation. Integrating lishers. Sunderland, Massachusetts. ef¤cient grassland Farming and Biodi- García MB. 2008. Life history and population versity. Proceedings of the 13th Interna- size variability in a relict plant. Different tional Occasional Symposium of the Euro- routes towards long-term persistence. Di- pean Grassland Federation Tartu, Estonia versity and distributions 14:106–113. 29–31 August 2005. (Eds. R. Lillak, R. Gärdenfors U. (ed.) 2005. Rödlistade arter i Viirtalt, A. Linke, & V. Geherman). Sverige 2005 – The 2005 Red List of Swe- Pärtel M, Mändla R, Zobel M. 1999. Land- dish Species. ArtDatabanken, SLU, Upp- scape history of a calcareous (alvar) grass- sala. land in Hanila, western Estonia, during the Gustavsson E, Lennartsson T, Emanuelsson last three hundred years. Landscape Eco- M. 2007. Land use more than 200 years ago logy 14:187–196. explains current grassland plant diversity Poschlod P, Bakker JP, Kahmen S.2005. in a Swedish agricultural landscape. Bio- Changing land use and its impact on biodi- logical Conservation 138:47–59. versity. Basic and Applied Ecology 6:93–98. Hodgson JG, Grime JP, Wilson PJ, Thompson Preston CD, Pearman DA, Dines T. 2002. New K, Band SR. 2005. The impacts of agricul- atlas of the British and Irish ¥ora. Oxford: tural change (1963–2003) on the grassland Oxford University Press. ¥ora of Central England: processes and pro- Ratcliffe DA. 1984. Post-Medieval and Recent spects. Basic and Applied Ecology 6:107– Changes in British Vegetation: the Culmi- 118. nation of Human In¥uence. New Phytol. Honnay O, Bossuyt B, Jacquemyn H, 98:73–100. Shimono A, Uchiyama K. 2008. Can a seed Reed DH, Frankham R. 2003. Correlation be- bank maintain the genetic variation in the tween Fitness and Genetic Diversity. Con- above ground plant population? Oikos servation Biology 17:230–237. 117:1–5. Richards JF. 1990. Land Transformation. In: Honnay O, Jacquemyn H. 2007. Suscept- The Earth As Transformed by Human Ac- ibility of Common and Rare Plant Species to tion. Global and Regional Changes in the the Genetic Consequences of Habitat Frag- Biosphere over the Past 300 Years. Ed. B. L. mentation. Conservation Biology 21: 823– Turner II) pp 164–178. Cambridge Univer- 831. sity Press with Clark University. Cam- Houle D. 1989. The Maintenance of Polygenic bridge. Variance in Finite Populations. Evolution Saccheri IJ, Nichols RA, Brakeiled PM. 2006. 43:1767–1780. Morphological differentiation following ex- Ingelög T, Andersson R, Thernborg M. (eds.) perimental bottlenecks in the butter¥y Bi- 1993. Red Data Book of the Baltic Region. cyclus anynana (Nymphalidae). Biological Part 1. Lists of threatened vascular plants Journal of the Linnean Society 89:107–115. and vertebrates. Swedish Threatened Spe- Schleuning M, Niggemann M, Becker U, cies Unit, Uppsala, in cooperation with the Matthies D. 2009. Negative effects of habi- Institute of Biologya, Riga. tat degradation and fragmentation on the Krach B, Krach JE. 1991. Muß das mittel- declining grassland plant Trifolium monta- fränkische Steppengreiskraut aussterben? num. Basic and Applied Ecology 10:61–69. Ber. Bayer. Bot. Ges.62:181–182. Smith UK. 1979. Senecio integrifolius (L.) Luijten SH, Dierick A, Gerard J, Oostermeijer Clairv. (Senecio campestris (Retz.) DC. B, Raijmann LEL, Den Nijs HCM. 2000. Journal of Ecology 67:1109–1124. Population Size, Genetic Variance and Re- Soejima A, Maik M, Ueda K. 1998. Genetic productive Success in a Rapidly Declining, variation in relic and isolated populations Self-Incompatible Perennial (Arnica mon- of Chionanthus retusus (Oleaceae) of Tsu- tana) in the Netherlands. Conservation shima Island and the Tôno region, Japan. Biology 14:1776–1787. Genes Genet. Syst. 73:29–37. Mathiasen P, Rovere AE, Premoli AC.2007. Spitze K. 1993. Population structure in Genetic Structure and Early Effects of In- Daphnia obtusa: Quantitative genetic and breeding in Fragmented Temperate Forests allozymic variation. Genetics 135:167–374. of a Self-Incompatible Tree, Embothrium Tasser E, Walde J, Tappeiner U, Teutsch A, coccineum. Conservation Biology 21:232– Noggler W. 2007. Land-use change and na- 240. tural reforestation in the Eastern Central Pärtel M, Bruun HH, Sammul M. 2005. Bio- Alps. Agriculture, Ecosystems and Environ-

91 Negative association between population size and quantitative genetic variance

ment 118:115–129. Widén B. 1987. Population biology of Senecio Tilman D, May RM, Lehman CL, Nowak MA. integrifolius (Compositae), a rare plant in 1994. Habitat destruction and the extinc- Sweden. Nordic Journal of Botany 7: 686– tion debt. Nature 371:65–66. 704. Turelli M. 1977. Random Environments and Widén B. 1991. Phenotypic selection on ¥ow- Stochastic Calculus. Theoretical Population ering phenology in Senecio integrifolius, a Biology 12:140–178. perennial herb. Oikos 61:205–215. Van Buskirk J, Willi Y. 2006. The Change in Widén B. 1993. Demographic and genetic ef- Quantitative Genetic Variance with In- fects on reproduction as related to popula- breeding. Evolution 60:2428–2434. tion size in a rare, perennial herb, Senecio Van Heerwaarden B, Willi Y, Kristensen TN, integrifolius (Asteraceae). Biological Jour- Hoffmann AA. 2008. Population Bottle- nal of the Linnean Society 50:179–195. necks Increase Additive Genetic Variance Widén B, Andersson S.1993. Quantitative But Do Not Break a Selection Limit in Rain genetics of life-history and morphology in a Forest Drosophila. Genetics 179:2135– rare plant, Senecio integrifolius. Heredity 2146. 70: 503–514. Van Rossum F, De Sousa SC, Triest L. 2004. Widén B, Wetterin M. 1999. Åtgärdsprogram Genetic consequences of habitat fragmenta- för bevarande av fältnocka (Tephroseris in- tion in an agricultural landscape on the tegrifolia). Stockholm: Naturvårdsverket. common Primula veris and comparison Willi Y, Van Buskirk J, Hoffmann AA. 2006. with its rare congener, P. vulgaris. Conser- Limits to the Adaptive Potential of Small vation Genetics 5:231–245. Populations. Annu. Rev. Ecol. Evol. Syst. Vindens Y, Engen S, Sæther B. 2008. Individ- 37:433–58. ual Heterogeneity in Vital Parameters and Willis JH, Orr HA. 1992. Increased heritable Demographic Stochasticity. The American variation following population bottlenecks: Naturalist 171:455–467. the role of dominance. Evolution 47:949– Waldmann P. 2001. Additive and non-additive 957. genetic architecture of two different-sized Wright S.1931. Evolution in Mendelian Popu- populations of Scabiosa canescens. Hered- lations. Genetics 16:97–159. ity 86:648–657. Young AG, Brown, AHD, Zich FA. 1999. Gen- Widén B. 1986. Fältnockan. En hotad art i etic Structure of Fragmented Populations Sverige. Rapport från projektet Artbeva- of the Endangered Daisy Rutidosis leptorr- rande – växter. Naturvårdsverket. Rapport hynchoides. Conservation Biology 13:256– 3055. 265.

92 Doctoral theses published at the Department of Ecology, Plant Ecology and Systematics, Lund University

1. Åke Persson (1962) Mire and spring vegetation in an area north of lake Torneträsk, Torne Lappmark,Sweden. 2. Nils Malmer (1962) Studies on mire vegetation in the archaean area of southwestern Götaland (South Sweden). 3. Sven Olov Strandhede (1966) Studies in European Eleocharis subser. Palustres. 4. Gertrud Nordborg-Dahlgren (1967) The genus Sanguisorba, sect. Poterium. Experimental studies and taxonomy. 5. Sven Snogerup (1967) Studies in the Aegean Flora VIII and IX. Erysimum sect. Cheiranthus. 6. Ingemar Björkqvist (1968) Studies in Alisma L. 7. Bertil Nordenstam (1968) The genus Euryops, morphology, cytology and taxonomy. 8. Tore Mörnsjö (1969) Peatland studies in Scania, South Sweden. 9. Arne Strid (1970) Biosystematic and cytological studies in the Nigella arvensis complex. 10. Folke Andersson (1970) An ecosystem approach to vegetation, environment and organic matter of a mixed woodland and meadow area. 11. Mats Sonesson (1970) Ecological studies on poor mire vegetation in the Torneträsk area, Northern Sweden. 12. Bengt Nihlgård (1970) Comparative studies on beech and planted spruce forest ecosystems in Southern Sweden. 13. Gunnar Weimark (1971) Studies of Hierocloe (Gramineae). 14. Germund Tyler (1971) Studies in the ecology of Baltic seashore meadows. 15. Bengt Bentzer (1973) Biosystematic studies in the genus LeopoldiaPoul. (Liliaceae). 16. Roland von Bothmer (1973) Biosystematic studies in Allium. 17. Alf Oredsson (1973) Investigation of the frequencies of Blackberry species in Sweden. 18. Ulf Olsson (1974) Biosystematic studies on oaks within the distribution area of Quercus petraea. 19. Karin Persson (1974) Biosystematic studies in the Artemisia maritima complex in Europe. 20. Hans Olsson (1974) Studies on South Swedish sand vegetation. 21. Lars Påhlsson (1974) Vegetation, microclimate and soil on slopes of some Scanian hills, southern Sweden. 22. Mats Gustafsson (1975) Evolutionary trends in the Atriplex prostata group of Scandinavia. 23. Torgny von Wachenfeldt (1975) Marine bentic algae and the environment in the Öresund. 24. Carl-Erik Nylander (1975) Vegetationshistoria och vegetation i södra Bräkne-Hoby, Blekinge. 25. Karin Brunsberg (1976) Biosystematic studies in the Lathyrus pratensiscomplex. 26. Lennart Engstrand (1977) Biosystematics and taxonomy in Geocaryum Cosson (Umbelliferae). 27. Lars Edler (1977) Phytoplankton and primary production in the Sound. 28. Ingvar Nilsson (1977) Biogeochemical cycling and net primary production in forest ecosystems. 29. Ewa Kvillner (1978) On the use of principal components and factor analysis in ecology. 30. Sven Jensén (1978) Sampling methods and numerical treatments applied to a classi¤cation of lakes in Southern Sweden based on macrophyte composition. 31. Ingvar Kärnefelt (1979) The brown fruticose species of Cetraria. 32. Håkan Staaf (1979) Decomposition and plant nutrient release from litter in two forest ecosystems. 33. Carin Tyler (1979) Studies on Schoenus vegetation in South and Southeast Sweden. 34. Bo Wallén (1980) Biomass, productivity and soil formation in early primary succession on sand dunes, S. Sweden. 35. Allan Nicklasson (1980) Vattenståndets och markanvändningens in¥ytande på strand- och bottenvegetationen i sydsvenska oligotrofa sjöar. 36. Lage Bringmark (1980) Processes controlling solutes in a pine forest soil. 37. Anna-Maj Balsberg (1980) Studies in a Filipendula ulmaria L. meadow ecosystem and the effects of cadmium. 38. Dagmar Persson (1981) Biosystematics of Stachys swainsonii Benth. (Lamiaceae) and its relations to some other chasmophytic Stachys species.

93 39. Magnus Magnusson (1981) Composition and successions of bryophytes and lichens in a coastal dune area in southern Sweden. 40. Thomas Karlssson (1982) Taxonomy and phytogeography of Euphrasia in Sweden, especially E. rostkoviana. 41. Ruth-Aimée Kornfelt (1982) Distribution and biomass of sublittoral macroalgae of Kullen, southern Sweden, correlated to some ecological factors. 42. Björn Widén (1982) Reproductive biology in the Helianthemum oelandicum(Cistaceae) complex on Öland, Sweden. 43. Ingrid Stjernquist (1982) Photosynthesis, growth and competitive ability of some coniferous forest mosses and the in¥uence of herbicides and heavy metals (Cu, Zn). 44. Stefan Persson (1982) Changes in vegetation, populations and environment during spontaneous successions in two plant communities in south Sweden. 45. Staffan Karlsson (1982) Ecology of a deciduous and an evergreen dwarf shrub: Vaccinium uliginosum and Vaccinium vitis-idaea in subarctic Fennoscandia. 46. Kristina Sundbäck (1983) Microphytobenthos on sand in shallow brackish water, Öresund, Sweden. 47. Britt Snogerup (1983) The genus Odontites (Scrophulariaceae) in north-west Europe. 48. Lennart Folkeson (1983) Heavy-metal pollution of forest ecosystems: effects on vegetation and mineralization of organic matter. 49. Gunilla Olsson (1984) Old-¤eld forest succession in the Swedish west coast archipelago. 50. Urban Emanuelsson (1984) Ecological effects of grazing and trampling on mountain vegetation in northern Sweden. 51. Bo Wiman (1985) Aerosol dynamics in coniferous forests empirical and theoretical analyses. 52. Björn Rörslett (1985) Regulation impact on submerged macrophyte communities in some Norwegian lakes. 53. Bo Sundström (1986) The marine diatom genus Rhizosoleni. 54. Gösta Regnéll (1986) Fen grassland in southern Sweden – the application of plant ecology to a problem of nature conservancy. 55. Sture Wijk (1986) Salix herbacea and the alpine snow-bed environment. 56. Bo Bergkvist (1986) Metal ¥uxes in spruce and beech forest ecosystems of South Sweden. 57. Hilde Nybom (1987) Apomixis in the genus Rubus and its effects on reproduction. 58. Roy Franzén (1987) Biosystematics of the Achillea clavennae and A. ageratifolia groups (Asteraceae). 59. Annette Carlström (1987) The ¥ora and phytogeography of Rhodos, Simi, Tilos and the Marmaris peninsula (SE Greece, SW Turkey). 60. Jan Thomas Johansson (1987) Taxonomic and morphological studies in the tribe Morindae (Rubiaceae). 61. Göran Svensson (1987) Studies on fossil plant communities, stratigraphy and development of peatlands in South Sweden. 62. Brita Svensson (1987) Studies of the metapopulation dynamics of Lycopodium annotium and its microenvironment. 63. Poul Hansen (1988) Statistical modeling of soil-macrofungal realtionships in south Swedish beech forests. 64. Lars Fröberg (1989) The calcicolous lichens on the Great Alvar of Öland, Sweden. 65. Thomas Landström (1989) The species of Ornithogalum L. subgenus Ornithogalum (Hyacinthaceae) in Greece. 66. Ingibjörg Svala Jonsdottir (1989) The population dynamics, intraclonal physiology and grazing tolerance of Carex bigelowii. 67. Ursula Falkengren-Grerup (1989) Soil acidi¤cation and vegetation changes in South Swedish forests. 68. Linus Svensson (1990) Flora morphology and the effects of crossing distance in Scleranthus. 69. Stefan Andersson (1990) Population differentiation in Crepis tectorum (Asteraceae): past and current patterns of natural selection. 70. Bengt L. Carlsson (1990) Controls on the growth and population dynamics of Carexbigelowii. 71. Dan Berggren (1990) Species of Al, Cd, and Cu in forest soil solutions – analytical methods, mobilization mechanisms, and toxicity to plants. 72. Ullmer Norden (1992) Soil acidi¤cation and element ¥uxes as related to tree species in deciduous forests of south Sweden. 73. Jan-Erik Mattsson (1993) The genus Vulpicida.

94 74. Maud E. Andersson (1993) Aluminium and hydrogen ions – limiting factors for growth and distribution of beech forest plants. 75. Mostafa Asadi (1994) The genus Elymus L. (Poaceae) in Iran: Biosystematic studies and generic delimitation. 76. Nils Cronberg (1994) Genetic diversity and reproduction in Sphagnum (Bryophyta): Isozyme studies in S. capillifoliumand S. rubellum. 77. Torleif Bramryd (1994) Effects on growth and nutrition of sewage sludge application in acid pine forests (Pinus sylvestris, L.) in a temperature gradient in Sweden. 78. Kerstin L. Sonesson (1994) Regeneration ecology of oak, Quercus robur L. – in¥uence of cotyledons and soil type on growth and nutrient uptake in seedlings. 79. Jörg Brunet (1994) Importance of soil solution chemistry and land use to growth and distribution of four woodland grasses in south Sweden. 80. Sigurdur H. Magnusson (1994) Plant colonization of eroded areas in Iceland. 81. Ulrika Rosengren-Brinck (1994) The in¥uence of nitrogen on the nutrient status of Norway spruce (Picea abies L. Karst). 82. Martin Ljungström (1994) Beech (Fagus sylvatica) seedling growth and nutrition – effects of acid soils and liming. 83. Ulf Arup (1995) Littoral species of the lichen genus Caloplaca in North America. 84. Leif Jonsson (1995) Effects of restoration on wooded meadows in southeasternSweden. 85. Arne Thell (1996) Anatomy and taxonomy of cetarioid lichens. 86. Marie Widén (1996) Clonal structure and reproductive biology in the gynodioecious herb Glechoma hederacea L. Lamiaceae. 87. Stefan Ekman (1996) The corticolous and lignicolous species of Bacidia and Bacidina in North America. 88. Alex Haxeltine (1996) Modelling the vegetation of the Earth. 89. Christer Albinsson (1996) Vegetation structure and interactions on mires. 90. Helena Runyeon (1997) Variation in Silene vulgaris and S. uni¥ora (Caryophyllaceae): genetic diversity, gene ¥ow and habitat selection. 91. Louise Lindblom (1997) The genus Xanthoria (Fr.) Th. Fr. in North America. 92. Olle Johnsson (1998) Genetic variation, clonal diversity and breeding systems in sedges (Carex). 93. Magnus Thorén (1998) Resource economy of carnivorous plants: Interactions between prey capture and plant performance in three subarctic Pinguicula species. 94. Gudrun Berlin (1998) Semi-natural meadows in southern Sweden – changes over time and the relationship between nitrogen supply and management. 95. Carola Gehrke (1998) Effects of enhanced ultraviolet-B radiation on subarctic ecosystems. 96. Lena Ström (1998) Organic acids in root exudates and soil solutions. Importance to calcicole and calcifuge behaviour of plants. 97. Gabrielle Rosquist (1999) Genetic variation, polyploidy and hybridization in Scandinavian Anthericum ramosum and A. liliago (Anthericaceae). 98. Åsa Olsson (1999) Morphometric and molecular variation in the Nordic dogroses (Rosa Sec. Caninae, Rosaceae). 99. Anna-Carin Linusson (1999) Changes in plant community diversity and management effects in semi-natural meadows in southern Sweden. 100. Lars-Erik Williams (1999) Nutrient cycling in agroecosystems: nitrogen cycling in southern Sweden in the 1850s and two Tanzanian villages in the 1990s. 101. Annika Kruuse af Verchou (1999) Reproductive strategies and liming responses in forest ¤eld-layer ¥ora. 102. Patrik Waldmann (2000) Quantitative conservation genetics of the rare plants Scabiosa canescens (Dipsacaceae) and Silenediclinis (Caryophyllaceae). 103. Ann-Mari Fransson (2000) Soluble and plant available phosphorus in acid soils. 104. Tina D’Hertefeldt (2000) Physiological integration and morphological plasticity in extensive clonal plants. 105. Angelika Zohlen (2000) Iron nutrition dynamics. Differences between calcicole and calcifuge plants. 106. Stephen Sitch (2000) The role of vegetation dynamics in the control of atmospheric CO2 content. 107. Sharon Cowling (2000) Plant-Climate interactions over historical and geological time.

95 108. Carin Nilsson (2000) Hemiparasites in the Subarctic: Resource acquisition, growth and population dynamics. 109. Anna Maria Jönsson (2000) Bark lesions and sensitivity to frost in beech and Norway spruce. 110. Gunnar Thelin (2000) Nutrient imbalance in Norway spruce. 111. Jed Kaplan (2001) Geophysical application of vegetation modelling. 112. Marion Schöttelndreier (2001) Wild plants can improve their rhizosphere chemistry in acid soils. 113. Anna Joabsson (2001) Methane dynamics in northern wetlands: signi¤cance of vascular plants. 114. Ursula J. Malm (2001) Geographic differentiation and population history in Silene dioica and S. hifacensis: variation in chloroplast DNA and allozymes. 115. Magnus Olsson (2002) Uptake of and preference for nitrate, ammonium and amino acids by understory species in deciduous forests. 116. So¤e Wikberg (2002) Carex humilis – a caespitose clonal plant: ramet demography, ring formation, and community interactions. 117. Katarina Schiemann (2002) Genetic variation and population differentiation in the forest herb Lathyrus vernus (Fabaceae). 118. Torbjörn Tyler (2002) Geographic distribution of intra-speci¤c variation in widespread eurasian boreo-nemoral woodland herbs. 119. Helena Persson (2002) The spatial structure of genetic and morphometric variation in Corylus avellana (Betulaceae): pattern and scale. 120. Aparna Misra (2003) In¥uence of water conditions on growth and mineral nutrient uptake of native plants on calcareous soil. 121. Anna Maria Fosaa (2003) Mountain vegetation in the Faroe Islands in a climate change perspective. 122. Anders Jacobson (2003) Diversity and phylogeography in Alisma (Alismataceae), with emphasis on Northern European taxa. 123. Christer Kalén (2004) Forest development and interactions with large herbivores. 124. Igor Drobyshev (2004) Interactions between climate, natural disturbances, and regeneration in boreal and hemi-boreal forests. 125. Anna Hagen-Thorn (2004) Nutritional ecology of selected Scandinavian tree species with special emphasis on hardwoods. 126. Ulrika Jönsson (2004) Phytophthora and oak decline – impact on seedlings and mature trees in forests soils. 127. Katarina Månsson (2005) Plant-bacterial and plant-fungal competition for nitrogen and phosphorus. 128. Martin Westberg (2005) The lichen genus Candelariella in western North America. 129. Samuel Kiboi (2005) Male and female selective mechanisms, reproductive success and gene ¥ow. 130. Margaret Mollel (2005) Environmental effects on pollen performance: potential consequences on gene ¥ow. 131. Teklehaimanot Haileselassie (2005) Effects of environmental factors on maternal choice and gene dispersal in plants. 132. Hans Göransson (2006) The vertical distribution of roots, mycorrhizal mycelia and nutrient acquisition in mature forest trees. 133. Rayna Natcheva (2006) Evolutionary processes and hybridization within the peat mosses, Sphagnum. 134. Eva Månsby (2007) Geographic variation, hybridization and evolution in the bladder campions, Silene vulgaris and S. uni¥ora (Caryophyllaceae). 135. David Ståhlberg (2007) Systematics, phylogeography and polyploid evolution in the Dactylorhiza maculata complex (Orchidaceae). 136. Louise Hathaway (2007) Patterns of geographic variation in Silene section Elisanthe (Caryophyllaceae): hybridization and migrational history. 137. Pernilla Göransson (2007) Genetic adaptation to soil acidi¤cation in four grasses. 138. Frida Andreasson (2007) Nutrient and organic matter dynamics in beech forest ¥oors in relation to the presence of ground ¥ora. 139. Karin Valtinat (2007) Plant colonization of oak plantations – the interactive effects of local

96 environment and land-use history. 140. Dirk-Jan ten Brink (2007) The role of regeneration in plant niche differentiation and habitat specialization. 141. Jakob Sandberg (2008) Soil phosphorus – a multidimensional resource that plays an important role for grassland plant species richness. 142. Triin Reitalu (2008) Plant species diversity in semi-natural grasslands: effects of scale, landscape structure and habitat history. (Joint PhD project with the Department of Physical Geography and Ecosystems Analysis.) 143. Lotten Jönsson Johansson (2008) Semi-natural grasslands: landscape, history and plant species diversity. (Joint PhD project with the Department of Physical Geography and Ecosystems Analysis.) 144. Johanna Eneström (2008) Life-history traits and population differentiation in a clonal plant: implications for establishment, persistence and weediness. 145. So¤e Nordström (2008) Systematics of polyploid Dactylorhiza (Orchidaceae) – genetic diversity, phylogeography and evolution. 146. Maarten Ellmer (2009) Quantitative genetic variation in declining plant populations. 147. Kerstin Isaksson (2009) Investigating genetic factors behind the decline of a threatened plant species – Tephroseris integrifolia (Asteraceae)