Sett elg og sett hjort-overvåkingen: Styrker og forbedringspotensial

Abstract

Målrettet hjorteviltforvaltning krever innsamling av fortløpende informasjon om bestandsutviklingen i kombinasjon med effektiv høsting. I denne prosessen har fritidsjegere en viktig rolle ved at de hvert år rapporterer jaktinnsatsen og antallet elg eller hjort de observerer under jakta. Disse observasjonene, kalt sett hjort- og sett elg-data, blir siden bearbeidet til relevante bestandsindekser og benyttes som beslutningsunderlag i den lokale viltforvaltningen. Til tross for utstrakt bruk, vet vi fortsatt lite om hvor presist sett elg- og sett hjort-indeksene er i stand til å reflektere endringene i de relevante bestandsegenskapene. Her rapporterer vi resultatene fra et 2-årig prosjekt der vi har evaluert bruken av bestandsindeksene fra sett elg og sett hjort-overvåkingen. Hovedmålet har vært å undersøke i hvilken grad varierende presisjon legger begrensninger på bruken av indeksene i lokalforvaltningen, og hvilke faktorer som best kan forklare manglende presisjon. Vi har spesielt undersøkt sannsynligheten for at viktige antagelser bak bruken av sett dyr-indeksene – som stabil oppdagbarhet og fordeling av dyr i tid og rom – er innfridd. I tillegg har vi utredet hvorvidt vi med små endringer i innsamlingsrutiner eller påfølgende analyser kan gjøre bestandsindeksene mer presise. Vi benyttet sett dyr- og aldersdata fra 16 ulike elgområder og 3 hjorteområder innsamlet over 7 (2006–2012) til 44 år (1967–2012). Ved bruk av kohortanalyse av kjønns- og aldersdata fra skutte individer rekonstruerte vi så bestandsstørrelsen og -strukturen i de ulike områdene, hvorpå vi undersøkte graden av samvariasjon mellom sett dyr-indeksene (sett pr. jegerdag, skutt pr. jegerdag, sett pr. km2, skutt pr. km2, kalv pr. hunndyr, hunndyr pr. hanndyr, spissbukk pr. bukk) og de relevante bestandsegenskapene (bestandstetthet, kjønnsrate og rekrutteringsrate) i studieperioden. I tillegg analyserte vi sett dyr-data innsamlet på dag- og jaktfeltnivå innenfor et utvalg av kommuner og år for å avdekke i hvilken grad antagelsene bak metoden var innfridd. Samlet sett fant vi at sett dyr-indeksene er i stand til å avspeile mellomårsvariasjonen i de ulike bestandsegenskapene, men med ulik presisjon avhengig av art, indeks og område. Vi fant høyere presisjon for indekser basert på sett elg-data enn på sett hjort-data, og høyere presisjon for tetthetsindekser (sett dyr pr. jegerdag, skutt dyr pr. jegerdag, antall dyr sett, antall dyr skutt) enn indekser på kjønnsrate (hunndyr pr. hanndyr) og rekrutteringsrate (kalv pr. hunndyr). Særlig høy samvariasjon fant vi mellom den rekonstruerte bestandsstørrelsen og antallet elg sett eller skutt pr. km2, mens antallet elg sett og skutt pr. jegerdag viste noe mindre samvariasjon. Det siste skyldes at antallet dyr skutt eller sett ikke øker proporsjonalt med jaktinnsatsen (antall jegerdager) og at jaktinnsatsen har variert mye i mange områder. I områder med stor variasjon i jaktinnsatsen er det sannsynlig at veksten i bestandstetthet underestimeres av sett pr. jegerdag-indeksen når jaktinnsatsen øker og overestimeres når jaktinnsatsen synker. Årsaken til at antallet observasjoner ikke øker i takt med jaktinnsatsen tror vi skyldes to forhold: 1) at mer marginale jaktområder (poster) tas i bruk med økende antall jegere, og 2) at elgene som observeres oftere kanselleres som dobbeltobservasjoner når antallet jegere pr. jaktlag er høyere. Førstnevnte forhold lar seg vanskelig kontrollere for ettersom forskjellene i jaktområdenes (postenes) kvalitet varierer mellom områder og sannsynligvis over tid. Effekten av det andre forholdet kan sannsynligvis reduseres vesentlig ved å endre noe på instruksen for rapportering av sett dyr. Alces alces, Cervus elaphus, evaluering, forvaltning, jakt, Norge, overvåking, sett elg, sett hjort, evaluation, hunting, management, monitoring, Norway, seen moose, seen red deer

Effective wildlife management often requires information on population density and performance of game species. In Norway, moose and red deer hunters have for many years reported their hunting effort, as well number, sex and age class of moose or red deer observed and killed. These observations, often called seen moose and seen red deer data (collectively: seen deer data), are then used to calculate indices of population density (deer seen per hunterday, deer killed per hunterday), recruitment rate (calves per female, twinning rate) and sex ratio (adult females per male). Although widely used by the local management to determine the annual size and structure of harvest quotas, we lack a good understanding of how precisely these indices predict the variation in relevant population characteristics. Particularly, we are concerned that variation in detection probability (i.e. detectability) and spatio-temporal variation in animal distribution affects their precision. Here we report the results from a 2-year project where we evaluated the precision of such population indices. The main aim was to assess the extent to which low or varying precision restricts the usefulness of the indices, and, if possible, to determine which factors best explain the lack of precision. To assess the precision we used seen deer and age-at-kill data from 16 moose areas and three red deer areas collected over 7 (2006–2012) to 44 years (1967–2012). We then reconstructed the population size and structure by the use of cohort analysis and age-at-kill data, and compared times series of seen deer indices with the time series of relevant population characteristics (size, sex ratio, recruitment rate) in the study areas. In addition, we analysed the spatio-temporal variation in seen deer indices based on data collected at the scale of day and hunting field in a subset of study areas and years. In general, we found that seen deer indices are able to reflect the annual variation in different population characteristics, but with varying precision depending on deer species, index and area. Overall, we found indices based on seen moose data to be more precise than indices based on seen red deer data, and density indices to be more precise than indices of sex ratio and recruitment rate. We found a particularly high correlation between the reconstructed population size and the number of moose seen or killed per km2, whereas the number of moose seen or killed per hunterday was slightly less precise. This is because the number of moose seen or killed did not increase proportionally with the hunting effort, and because the hunting effort showed large annual variation in many areas. In such areas, it is likely that the number of moose seen and killed per hunterday under-estimate the population growth rate in years of increasing effort and over-estimate the growth rate in years of decreasing hunting effort. The lack of a proportional relationship between number of observations and effort is most likely related to two conditions: 1) on average more marginal hunting areas are used as the hunting effort increase, and 2) relatively more observations are cancelled as double observations as the number of hunters per team increases. The former condition is difficult to control for as the effect of hunting effort on detectability seems to vary among areas and probably also over time. The effect of the latter condition, however, can be significantly reduced by making only small changes in the instruction for how to record and report moose and deer observations. Analyzing the subset of data collected at the level of day and hunting field, we found that the detectability of moose and deer and their spatial distribution are not uniform throughout the hunting period and across years. This was particularly evident in the red deer areas where the number of red deer seen per hunterday increased during the hunting season in two out of three populations, despite a significant reduction in population size. We believe this to occur because red deer concentrate in smaller areas at lower altitude during the autumn, which in turn increase the effective population density in the areas where they are actually hunted. Annual variation in the timing of such concentrations can partly explain the poor correlation between number of red deer seen per hunter day and population size across years. We also suspect that hunting effort that does not lead to observations of deer is less likely to be reported in red deer areas. Such a violation of instructions will decrease index precision and reduce our abilities to detect population declines. Also the observed sex ratio and recruitment rate changed in accordance with the hunting regime in the moose areas. E.g. the proportion of observed adult females increases during the hunting season, in accordance with the higher proportion of adult males and calves that were harvested. A similar pattern was not found in red deer areas, which indicates that the detectability of males, females and calves change with different rates during the hunting season. Possibly, this is related to the smaller size of red deer than moose and the fact that the red deer hunting season lasts three time longer (15 weeks) than the moose season. During this period red deer migrate from summer to wintering areas, rut, and experience significantly changes in the environment, all of which may affect the behavior of the different sex and ageclasses. To the extent this behavior is also affecting their detectability, annual variation in phenology will have a significant effect on the precision of the various seen deer indices across years. In general, we believe that the indices from the seen moose and seen red deer monitoring have much value for the local wildlife management. The seen deer monitoring is relatively cheap to conduct on a regular basis, and it survey the population size when moose and deer are in the hunting areas (summering areas), i.e., where they are most actively managed (hunted). Alternative census methods, potentially with higher accuracy, are significantly more expensive and are preferably conducted in winter when snow covers the ground. At that time, however, the population may have significantly changed due to migration, and hence may not reflect the populations that are managed during the autumn hunting season. We therefore conclude that the seen moose monitoring provides a good alternative to other census methods in the moose areas, and that the seen moose indices in most cases will reflect the correct population development. The precision of the red deer population indices are less precise and therefore we advise the management to be more cautious in predicting the population development based on these indices. Nevertheless, the seen red deer monitoring should be continued. The seen red deer monitoring is still in its infancy compared to the seen moose monitoring and so far we have little experience in how to collect, process and interpret seen red deer data. Before concluding with regards to its usefulness, we therefore need more years of seen red deer data of high quality.

At the end of the report we list several suggestions on how to improve the process of collecting, reporting and quality assuring seen deer data, and how the local and central management institutions can utilize the above results. At the local level, we particularly advise the management to include information about the hunting effort when they interpret the variation in seen and killed moose per hunterday as an index of density. If hunting effort increase substantially, the population growth rate is likely to be underestimated. For the same reason the Norwegian Environment Agency should consider a change in the instruction for how to collect and report seen deer data. In its current form hunters are asked to cancel all observation of moose or red deer that with some certainty have been seen previously in the same day by any of the team members. In our view, it is sufficient to cancel doubleobservations of the same animal in the same hunting situation, but not observations of the same individual by other team members or in other hunting situations. To improve the seen deer monitoring we need more research into the factors leading to variation in detectability of moose and red deer. The increasing use of GPS-radiocollars are now providing detailed information about the behaviour and movement pattern of moose and deer, but only in few cases is this information used to learn more about their detectability during the hunting season. By combining data on deer and hunter behaviour in the same areas, we can learn more about the factors making some categories more detectable than others and why it varies over time. In addition, such studies can provide more information about the temporal variation in moose and red deer distribution during the hunting season, and to what extent this varies between years. In particular, it is important to determine how concentrations and expansions of animals affect the hunting activity and number of deer observed. Based on experiences in fishery research, the detectability (or catchability) seems first of all to be affected by fishing effort and methods. Accordingly, we need more information about the various methods used for hunting moose and red deer throughout the country, and how this is likely to affect the relative number of moose and deer observed.