Vargjakt effekter Laikre 2013

12 Feb

Hunting does not increase inbreeding in Swedish wolf, in spite of that this was suggested in a recent publication

Summary

Laikre et al 2013 find that the hunt 2010 significantly increased inbreeding in the culled population. There are other explanations of the observation, which seem much more likely, but the authors have neither discussed nor mentioned that. It was not tested if an equivalent material from an equivalent hunt 2011 gave the same result. The hunt 2010 actually very likely reduced the future inbreeding. The major result is thus not supported by data. It also contradicts a later study based on more and better data. Thus the trustworthiness of the results and conclusions are low.

The paper also expresses the view that 1500 wolves are needed as Swedens contribution to a sufficient large ”metapopulation”.  I do not think the need of a large genetically connected metapopulation is a problem with a small Swedish contribution, see a newer page.

 

Bakgrund:

Jag har uppmärksammats på en uppsats, som bifogats de ideella naturvårdsorganisationernas jaktöverklagande. Jag bads om en bedömning ur jaktsynpunkt. Jag skriver en utvärdering på engelska. Det är naturligtvis viktigt att den här bedömningen inte baseras på missuppfattningar eller fel av mig och jag ber därför alla som vill påpeka möjliga misstag i värderingen att kommentera. Uppsatsen har mycket beskrivning och diskussion om vargpolitik och detta diskuteras mycket i andra sammanhang i Sverige, och det urval av argument som presenteraras i uppsatsen är troligen väl kända. Jag fokuserar helt på de originaldata som presenteras och analyseras, som leder till slutsatser om vargjaktens effekter, som kan ha relevans för den juridiska bedömningen av selektiv vargjakt. Huvudresultaten och påståendena i uppsatsen har sedan mer än ett år presenterats för bland annat naturvårdsverket och vid seminarier som är tillgängliga på webben och de har presenterats av Dagens eko i intervjuer. De har lagts fram vid ett licentiat-seminarium. Studien tillför alltså inte genuint ny information till den svenska vargdebatten, men resultaten dokumenteras några månader senare i en sådan form att det vetenskapliga underlaget kan granskas.

Evaluation of results and conclusions

Linda Laikre, Mija Jansson, Fred W. Allendorf, Sven Jakobsson and Nils Ryman: Hunting Effects on Favourable Conservation Status of Highly Inbred Swedish Wolves, Conservation Biology 2013.  DOI: 10.1111/j.1523-1739.2012.01965.x

What is written in abstract about hunting is ”Hunting to reduce wolf numbers in Sweden is currently not in line with national and EU policy agreements and will make genetically based FCS criteria less achievable for this species”.

This is very severe accusations when not formulated by reservations like ”The authors suggest that…”. I use different colors for two different statements above. The Government of Sweden does not share the opinion of the authors. It is remarkable and should not be tolerated by a self-respecting scientific journal that such statements are made as a matter of fact.  This is not at all supported by the data presented in the paper (see below) neither it is supported by other studies. The Swedish government attached an evaluation of the genetic effects of the actual hunt in a response to EUs legal opinions, which does not support the analyse of the hunt of the current paper. This study is not mentioned. 

The paper is not arranged in the common format for a scientific paper with sections for material and methods, results and discussion. This would have been beneficial as it forces the author to discuss the results and it makes it more evident what is political discussion and what are the actual results.

The link between data and results
The experimental results (DNA-analyses and interpretations from them, the main bulk of the work behind the study) are made by M. Åkesson and others at Grimsö. What is written about the material is that it was obtained by Skandulv. It raises some suspicion that Åkesson is neither in the author list nor mentioned in acknowledgements, he may not have agreed on statements in the paper or conclusions of the results. Neither Skandulv is mentioned in acknowledgements. It is unfortunate for the study that the competence familiar with the data and their limitations has not taken part in the evaluation of the data, the intellectual distance between the data and their evaluation has contributed to severe mistakes in the interpretations. Also the paper reevaluates the same data in a worse way than an evaluation presented some years ago without mentioning it, this would not have been done by a closer contact with the data supplier   It is generally a good advice for scientific studies to be close to the data the results are based on. I have asked for Åkessons comment and have not got any objection against the descriptions in this web-article.

The study finds the coefficient of inbreeding is on average lower in the wolves culled in the hunt than in the total population. If there is a causal relationship, so the wolves with lower inbreeding are more likely to be culled at the wolf hunt, and that such small difference in inbreeding (F=0.03) would cause this is indeed extremely surprising and unexpected. Inbreeding differences of 3% very seldom cause a phenotypic difference of more than 3% (perhaps with an exception for fertility), and the change in likelihood of being culled to cause the suggested effect is much larger than that. A 3% difference in likelihood of culling among 28 culled would correspond to something like 0.3 wolves, much too small difference to be identified in statistical noise! No effort has been done to discuss an effect of inbreeding which could be sufficient large to cause the observed difference if it were a true causal relationship.

The study was not repeated in spite of that this had been very easy. If such highly surprising and contra-intuitive results should be believed, the authors should use all material easily available. In this case exactly the same material for the hunt early 2011 is also available and it would have been easy to repeat the same analyses. It would have increased the trustworthiness of the results if there was a replication. It decreases the trustworthiness of the study that the evident risk that the results should not be repeatable was not taken.

Significance may not mean that there is a relationship between factors which analytically appear as related. First it must be discussed if the relationship really is significant and that what appear as significance is just a random event. A significance test as performed may be OK for a well organised experiment, but for observations from non-experimental material alternative reasons for the ”significance” meaning no “true” meaningful relationship between inbreeding and chance to get killed must be discussed. No efforts to do that have been done in spite of a very wordy and detailed discussion on other matters.

The wolves exposed for hunting are not those assumed. The author’s base their conclusions that their result is significant on that there are exactly 195 wolves exposed for the hunt, all with a known F. That is not so, the number of wolves in Sweden available for hunting is highly uncertain, and the F values of the unknown wolves are of cause not known. Neither can it be inferred from inventories that all “existing wolves” with known F really were exposed to the hunt. A wolf can be identified a month before the hunt, but die before the hunt. Some wolfs (besides the protected) were in areas of Sweden where wolf-hunt was not permitted. A “Swedish” wolf can be in Norway at the day of the hunt or a “Norwegian” wolf can be in Sweden. If the uncertainty comprises 50 wolves and the standard deviation of F is 0.10, the 95% confidence interval of the average of the unknown wolves is in the same magnitude as the difference observed.

F-values for culled wolves are interfered from different sources than surviving. I am convinced the F based on DNA analyses from the killed and the remaining living origin from different sources (this is NOT described in “material and methods” NEITHER in discussion). DNA samples are taken from all killed wolves and a reliable F can be derived. For the non-hunted most F values are derived from assumptions on the genetic and geographic structure (what family group the wolves are assumed to belong and the assumed size of the family group). A considerable number of wolves do not belong to an identified family group and escape DNA estimate unless they are hunted. The total number of wolves is not known very accurate (+-20) and thus the wolves characterised as non -hunted are just a sample even if comprising most of the ”non-hunted” wolves. Conclusions based on information derived in different ways for the compared populations is uncertain and the statistical analyses performed does not consider that.

The observed effects may be caused by geographic differences.

Most wolves belong to family groups which occupy geographically different territories. Incest happens in wolf by full sib and parent – offspring mating. All wolves with F>0.4 origins from such incest mating and they form a distinct extreme tail in the distribution of F shown in the study (Fig 1 a). Almost all of the highly inbred wolves live in a few territories (family groups, geographic areas, territory, wolf packs). Successful hunt may not have occurred in these territories, because of specific circumstances like snow tracking conditions, efficiency of organisation of the hunt, fast success in neighboring areas stopping further hunting, land ownership, skills by local hunters and hunting leaders, geographic localisation of the wolf pack during the hunt (which typically lasted less than a day). The death statistics for the winter 09/10 is dominated by the hunt and more than half the mortality occurred in one fifth of the wolf packs. The chance of being culled at the hunt 2010 is certainly not random distributed over the population but is very different in different territories. Killed wolves can not be seen as individual independent samples of the wolves available for culling.  That no single of the most inbred wolves was culled can be explained by where the few relevant wolf packs happened to be at the time of the hunt, which has nothing with inbreeding to do.

There is no expectation that hunt according to the Swedish model would reduce inbreeding among the wolves available for hunting, the mechanism the hunt reduces inbreeding is that genetically valuable wolfs are not available for hunting. A variation in the outcome by chance is expected and the authors observation in no way exclude a chance event. To focus on that it is ”significant” is highly misleading. The chance variation may even be larger than the causal reduction of inbreeding by culling shadowing this effect. There is nothing surprising in that if it is not shown that it is a unlikely (significant).

A confidence estimate based on independence of the hunted is just not reliable, and a minimum requirement for trustworthiness should be that the possible consequences of that are discussed. The main result of the study is not supported by the data analyses, thus it could be neglected in future Swedish wolf hunt considerations.

The hunt 2010 actually reduced inbreeding!

What matters for future inbreeding is not average of individual inbreeding, but contribution to “population inbreeding”, the “average kinship” (the terminology varies) in coming generations. The study is well aware of this and recommends that culling should be based on that measure. Still the study does not point out the impact of the hunt 2010 using this superior measure of estimating “inbreeding” is presented before by the Grimsö group in the report from the Swedish Government to EU August 2011. The decrease in average kinship after the quota hunt 2010 is clearly identifiable in the results from Grimsö shown in the figure in this article. As far as I understand the hunt 2010 actually reduced inbreeding still more than visible in the figure (see below), even if the significance of that off course can be discussed.

This paper generally neglects facts and considerations which are favorable to hunting.

The incest wolves seem to contribute much less to the future inbreeding than other wolves. The Scandinavian incest progenies seem to have reproductive success reduced to a third compared to the more typical inbred wolves and will thus contribute little to future generations even if not culled. http://vargdag.wordpress.com/2011/11/12/stamtavla-och-resultat/   Average phenotype inbreeding over whole population did not decrease by the hunt 2010, neither did it rise. This reflects random stochastic variations in the outcome sparing the incest progenies, and is no contraindication that the theoretical expectation that selective culling reduce inbreeding is wrong.

The inbreeding reduction following the hunt 2010 is a logic necessity,  but its size is too low to be confirmed by statistically significant observations mainly because of the impossibility to accurately enough define a comparison ”non-hunted” group. The possibilities to verify by molecular data are improved in the 2011 hunt.

The paper argues for a still stricter selection of high F individuals for culling. But as the most inbred are the incest progenies, the actual long-term effect is not likely to be improved by such a selection scheme in the actual case as the authors suggest.

Genes are claimed to be better preserved in a growing population. But looking in a long time perspective it is not so, only short time!

The study claims that more wolfs have  more genes and wherefore valuable genes are lost when wolves are shot.
If wolves are removed from a population some genes are lost. But sooner or later the population become constant and when the rarer genes will be lost more seldom thus it is favorable long time even if not immediately.

As long as a population is isolated it more individuals means more genes, but if there come migrants, they get less impact the larger the population is. Thus it is more difficult to cure inbreeding and get more of new genes in a larger population.

Hunts 2011-2014 are expected to be more efficient in reducing inbreeding

2011, 2012 and 2013 progressively more wolves have become “genetically valuable” and constitute low F, and their protection will contribute more to reduced inbreeding than they did in the hunt 2010. And as far as I understand Sweden has followed the advice of the study to apply more sophisticated tools (“average kinship”) than outright phenotypic F which will improve the efficiency slightly. This concerns the few wolves culled in the hunt 2013, but because the low culling intensity (only 3 wolves) the impact will be small, too small for compensating for genetic drift, so my guess is that the inbreeding will start raising again 2013, but this hypothetically reversing trend can still be prevented by selective culling 2014.

Swedish report about the genetic effect of culling

Liberg & Sand (2012) presented a report at mid-December. Their finding was that directed culling can be very efficient in a program heading at reduced inbreeding http://www.naturvardsverket.se/upload/04_arbete_med_naturvard/jakt/Selektiv%20och%20riktad%20jakt/Selective%20harvest%20Swedsioh%20wolves%20Final%20report%2015%20Dec%202012.pdf .  The report covered culling operations in the actual population 2013 (as was initiated, but with pathetically low intensity) and 2014. It also covered a longer period with migration and found that culling keeping wolf population to some hundred rather than 500 was efficient in reducing inbreeding (and thus improving heterozygosity). The conclusions by Liberg & Sand 2013 are on the relevant points, where conclusions differ, better supported by relevant data than the considerations by Laikre et al 2013. The main methodological objection by Laikre et al 2013 is already considered in current wolf management, and that the scientific fine tuning was not perfect in end of 2009 when the first wolf hunt was planned and most wolf scientists have had not incitement to think along these paths seems irrelevant to dig up now as a reason for severe objections.

Other type of statements in the study

The study deals with outermost political touchy considerations in Swedish wolf politics. Its actual results are doubful and the presentation is more about giving statements detailed opinions about what decisions Sweden as a state ought to decide in a way so it could be claimed to be refereed by anonymous referees and thus highly reliable on the roles of the EU-directive. E.g. the study states that at least 1000 sexually mature (2 year old) wolves are needed in Sweden to fulfill the demands by EU-regulations, provided they are genetically well connected with other subpopulations, (I guess this corresponds to 1500 after winter culling). I do not review or comment this type of statements here.

The study was presented at a licentiate seminar

The study has been a part of the licentiate thesis by Mija Jansson. The abstract of the thesis is on http://www.popgen.su.se/Licentiatavhandling_MijaJansson_Abstract.pdf . For licentiate thesis at my former University the rule is “Public display of the thesis shall be done at least three weeks (within the semester) before…” I cannot see that this has been done for this thesis. But it is another University and I may not have looked or read careful enough or misunderstood. The basic idea of a public thesis presentation is that anyone can study it long before the public presentation and at the presentation (seminar for a licentiate thesis and defense for a doctoral thesis) orally raise “questions” which may refer to weaknesses. I strongly support this practice as it is the ultimate safeguard for quality standards, and encourage Stockholm University to make use of the practice. I focused the attention to the event to the group headed by Olof Liberg, who was preparing a report on the genetic effect of hunting at the time of the licentiate seminar.

Linda Laikre, Mija Jansson, Fred W. Allendorf, Sven Jakobsson and Nils Ryman: Hunting Effects on Favourable Conservation Status of Highly Inbred Swedish Wolves, Conservation Biology 2013, Article first published online: 2 JAN 2013 | DOI: 10.1111/j.1523-1739.2012.01965.x  

Annonser

2 svar to “Vargjakt effekter Laikre 2013”

  1. DagL 26 augusti, 2013 den 13:44 #

    Någon kritisk kommentar om denna granskning har inte mottagits till 140101, ingen läsare har hört av sig annat än för att tycka att utformingen kunde göras mer klatchig, men det vill jag inte göra eftersom det är en vetenskaplig granskning. Det vetenskapliga utlåtandet har alltså utsatts för offentlig granskning utan att kritiseras.

    En fördel med att Laikre och Ryman har publicerat sina åsikter tidigt 2013 inklusive det vetenskapliga underlaget för dem är att åsikterna och underlaget för dem är väl kända och väldokumenterade och alltså beaktade i det fortsatta arbetet utan att nya synpunkter behöver inhämtas.

    Synpunkterna har diskuterats vid ett expertmöte på Naturvårdsverket 26/4, dit Laikre och Ryman inbjudits, men tydligen inte varit närvarande.

Trackbacks/Pingbacks

  1. FLASH: Undermålig Vargjaktsrapport används i förvaltningsrätten.. | Cullingman - 12 februari, 2013

    […] Vad Genetikprofessor Dag Lindgren gör för bedömning framgår mer än tydligt i något som inte kan kallas annat än ett totalt sågande av en uppsats, som har bifogats trojkans överklagande av Naturvårdsverkets selektiva vargjakt 2013. Uppmanar här alla som följt turerna att läsa Dags review som lagts ut på https://vargweb.wordpress.com/2013/02/12/vargjakt-effekter-laikre-2013/. […]

Kommentera

Fyll i dina uppgifter nedan eller klicka på en ikon för att logga in:

WordPress.com Logo

Du kommenterar med ditt WordPress.com-konto. Logga ut / Ändra )

Twitter-bild

Du kommenterar med ditt Twitter-konto. Logga ut / Ändra )

Facebook-foto

Du kommenterar med ditt Facebook-konto. Logga ut / Ändra )

Google+ photo

Du kommenterar med ditt Google+-konto. Logga ut / Ändra )

Ansluter till %s