Kisemlős populációk paramétereinek becslése és modellezése

Horváth Győző
Kisemlős populációk paramétereinek becslése és modellezése.
PhD, University of Szeged.

PDF (disszertáció)
Download (12MB)
PDF (tézis)
Download (180kB)
PDF (tézis)
Download (177kB)
[img] PDF (bírálati lap)
Restricted to Repository staff only

Download (700kB)

Abstract in foreign language

This dissertation contains the results of small mammal studies performed at the University of Pécs and small mammal population level monitoring of distinguished habitats along Drava river with support from Duna-Drava National Park. Small mammal research in the Drava Lowlands had been started well before co-ordinated biodiversity monitoring was launched or Duna-Drava National Park was established: small mammals have been trapped in a variety of habitats since 1994. Our research history thus covers more than a decade, including data series spanning several years as well as samplings with data form shorter periods in habitats of the upper Drava reach. The present dissertation summarises the results of our small mammal investigations on the basis of two synbiological aspects (population biological and demographic changes, survival estimation and modelling). In the first subject we dealt with the population dynamics of common species, demographic changes and the estimation of population size. Secondly, in addition to analysing and estimating abundance and its changes over time, the dissertation deals also with survival estimation and modelling, which comprise one of the most speedily developing subjects of population biology. Part of this research can be regarded, most of all, as syn-phenobiological investigation, as we have revealed synbiological phenomena, and analysed the temporal-spatial changes of certain population biological characteristics. In several cases, however, we evaluated our data in respect of various background variables (e.g. weather factors, physiognomy of the vegetation in the analysed habitats), thus the revealed phenomena could be scrutinised in a causal interrelation. The synbiological analysis and monitoring of small mammals was performed between 1994-2007 in habitats of Baranya and Somogy counties, along river Drava. The first sample area was selected in 1994 in Baranya, in a lowland oak-hornbeam forest habitat where small mammal studies were then continued for 10 years. In 1997 we performed the small mammal fauna survey of a mosaic habitat beside the Matty lake in the lower Drava section, which was followed by the analysis of the spatial and temporal pattern of small mammals in this lakeside habitat with heterogeneous vegetation for another six years. As a separate programme beside regional and national level projects of the Hungarian National Biodiversity Monitoring System, the biological monitoring of the upper Drava reach was launched, as part of which we started the monitoring of small mammal populations in 2000, in additional habitats. Two sampling sites were selected in Lankóci-forest near Gyékényes, differing from each other in their vegetation structures (a strictly protected, closed alder gallery forest, and a neighbouring reforested area. Furthermore, trappings were performed in 2007 along the upper Drava reach in Croatia, where the small mammals of Repas forest, a lowland oak-ash-elm gallery forest located in a higher terrain than the floodplain, this habitat type representing at the same time areas under forestry management. The method applied in each of the sampling areas was capture-mark-recapture, with the same box-type live-traps (75x95x180 mm). Just like the traps themselves, the trapping technique was also alike in all cases: bacon and cereals mixed with aniseed extract and vegetable oil were used as bait. The traps were checked twice daily in each of the forest habitats (Bükkhát-forest, Lankóci-forest, Repas-forest): from 700 in the morning and from 1900 in the evening, with the traps being left triggered during the day, because they did not heat up considerably inside the forest. However, the traps were left open for the day hours in the summer period beside the Matty lake, thus we had data from the morning trap checking sessions that resulted from greater nocturnal small mammal activity. The same code table was used for marking in all the sample areas. A separate database system was developed under Microsoft Access for storing and processing multiple capture and recapture data. SQL filtering of capture-recapture data made it possible or us to organise and process our data in a versatile and effective way. Each record of the database corresponds with data of one captured individual (either marked or unmarked), and the structure of its attributes, fields and associations with other data tables can be continuously widened. The long-term temporal patterns of population dynamics were analysed in the case of yellow-necked wood mouse (A. flavicollis), sriped field mouse (A. agrarius) and bank vole (C. glareolus) populations, all being frequent species in the Bükkhát-forest. Our long-term trapping survey between 1994-2003 was suitable for the analysis of demographic fluctuations of these frequent species, for which purpose we used the “minimum number alive” (MNA) values standardised for 100 trap nights. Based on the monthly values of this population index, cyclicity was tested for with an auto-correlation method. The fluctuation revealed in the case of the bank vole showed a 3-year repetition of waves with large amplitude, suggesting quasi-cyclic population dynamics. Similar result was obtained for the yellow-necked wood mouse: three higher abundance peaks were recorded, forming after three and four years, respectively. However, in the case of the striped field mouse population, unlike in the two other species, regular oscillation was revealed in the majority of the analysed 10 year period, and this oscillation was continuously fading out. Thus, peculiar annual population dynamics were recorded with the abundance peak occurring in August-September. The auto-correlation analysis greatly supported what had been found out already about the striped field mouse population in that this species has yearly cycles with autumn peaks followed by a spring population collapse and than another cycle starting over with the peak setting in autumn. The winter population decline was determined on the basis of mean autumn and spring density values (seasonal mean MNA). In the case of yellow-necked wood mouse and striped field mouse the population was found to be declining in each winter period. In bank voles, however, four winter seasons yielded significant drop in numbers, mostly coinciding with severe winters, but there was one winter period when abundance was increasing instead of declining. The reason for population growth in the winter of 1996-1997 was that the climbing of small mammal density values started quite late in the season in the preceding year (1996). Accordingly, autumn maxima, too, formed later in the season, and remained lower than in the former years. The growth of density values late into the autumn was partly due to the favourable weather conditions in autumn 1996 (higher average temperature, rainfall, no frost in November. The significance of this irregular year was reflected also in the regression analysis of abundance values from the autumn and the successive spring. The hypothesis that autumn abundance values are crucial in forming abundance relations in the spring i.e. at the onset of the reproductive season had to be rejected in all sampling years. This is due to the atypical years with favourable autumn and mild winter weather, allowing reproduction late in the autumn or into the winter. Only when these atypical years were left out from our calculations, did we receive the expected linear correlation in the case of the yellow-necked wood mouse, i.e. that autumn population size determined spring densities. In the improvement of the statistical evaluation of capture-recapture data an important line is the development of the combination of sampling methods. The first of such intentions leading to this direction produced a scheme that combined open and closed population models within a single analysis. How to schedule sampling while carrying out the programmes of the National Biodiversity Monitoring System and while working out its monitoring protocols seemed to be a problematic issue in planning the population-level monitoring of small mammals using live trapping. Accordingly, we tested for the suitability of Pollock’s robust method, for which we used data of the three frequent species occurring in the Bükkhát-forest habitats (A. flavicollis, A. agrarius, C. glareolus). Our results proved to be in line with Pollock’s simulation method which indicated a greater difference of Jolly-Seber estimators. Based on our monitoring experience, it seems reasonable to select more than three primary periods. As to secondary periods, our experience have showed that five sampling nights per one period are essential for the analysis of small mammal capture-recapture data with robust statistical methods. In survival estimation and modelling, we first dealt with the long-term analysis of winter survival and its estimation with various methods. Two methods were used for estimating the survival of bank voles from capture data obtained in the Bükkhát forest habitat. First, the 28-day survival calculated on the basis of the exponential winter decrease of abundance, then we used life-history matrices produced from capture data from an eight-year period (1994-2001), to estimate survival using the MARK computer program. When calculating 28-day survival rates we looked at how the values depended on weather factors (average winter temperature, average winter precipitation, average snow depth), and on the biotic factors of the population (bodily condition of adult individuals, proportion of reproductive females in the autumn period). In addition to these we also looked at how effect did the autumn abundance of the population have on winter survival. Apart from the years with exceptional condition, positive linear correlation was found between the size of autumn population and abundance in spring, this finding supporting our hypothesis. The 28-day winter survival values calculated on the basis of exponential winter population loss was found to be in correlation with average winter temperature and average snow depth. The result of modelling performed using the MARK program was the same as what was revealed by performing the traditional way of winter survival calculation. Data matrices generated from yearly capture-recapture values were suitable for modelling the effects of the assumed limiting co-variables on winter survival. Model selection was done on the basis of AIC-values. Statistical results of model selection showed that the winter survival of bank voles was positively influenced by higher winter temperatures and by the presence of snow cover. Among the biotic variables, it was the mean body mass of adult individuals i.e. the physical condition of adults in the autumn, and the proportion of reproductive females at that time in the population that determined the estimated value of survival. Thus, our results from estimating and modelling winter survival proved our hypothesis we had raised in respect of the bank vole population. A turning-point in capture-recapture data evaluation was the retrospective analysis of capture history; based on this approach it became possible to estimate true population growth between study periods. Furthermore, this method allows the researcher to see at what degree the various demographic parameters actually contributed to the growth of the population. We used the Pradel model, having been developed based on this method, to estimate population increase and growth rate. We estimated these parameters, together with survival, in the case of the dominant rodent populations (yellow-necked wood mouse, striped field mouse, bank vole) of three different forest habitats selected in the Hungarian upper Drava reach. We intended to find out to what extent the various components of demographic parameters (population growth resulting from the adult individuals of the previous sampling periods and from immigrating specimens) actually contributed to the increment of the population size. Our results showed that it was population growth resulting from the survival of adults that had the greatest importance in forming the growth tendencies in dominant rodent species of various forest habitats. In respect of the studied reforested habitat we emphasised that the decline of striped field mouse population, a species preferring highly dense vegetation, was positively influenced by bank voles intensively spreading in the area. This finding on expansion dynamics was supported by our calculations of population growth rate: expansion was proved mostly by the high values obtained for new individuals appearing in the population. Accordingly, the considerable rate of bank vole population growth in the reforested habitat was mostly due to immigration and colonisation. This rate of population growth and the resulting dispersion and spatial translocation is specially notable with a view to the fact that striped the field mouse that was competed out in our case is known in small mammal ecology as an expansive species with rapid population growth capacities. For survival estimations, we finally applied metapopulation approaches of multistate models. We studied the space use of the European pine vole (M. subterraneus) in two landscape scales in a mosaic habitat complex besides the Matty lake, on the basis of a five-year trapping period. First we considered the 50x50 m sample areas as habitat islands, then looked at the habitat selection of the same species on a finer, microhabitat scale. As well as for patch preference calculations, we considered the mapped microhabitat patches for survival estimations, too. From modelling survival and movement probabilities between the grids representing habitat islands it could be concluded that one of the most important factor throughout the four sampling periods was the time dependence of estimated parameters (survival probability, movement probability between sampling sites or habitats). Among the constraint factors and effective factors incorporated in this global model it was distance that played a more important role rather than the quality of habitats representing the most suitable patch mosaic for the species. At a microhabitat scale we could indicate that the most suitable microhabitat for pine voles was the patch dominated by Galium, as supported by data from all three grids. This species showed positive preference for Solidago type (in two of the grids) and Typha type (in one grid) patches as well. From the results of patch preference calculations we concluded that the pine vole reacts to the alteration of microhabitat quality by the spatial rearrangement of individuals, i.e. it shifts to the use of optimal patches at the finer spatial scale, thus presenting a coarse grained response to habitat heterogeneity. The patch-level multistate modelling of microhabitats further supported our conclusions specified above, in that the pine vole highly prefers patches with closed vegetation type whose quality is an important ecological constraint in forming the spatial distribution patterns of the population.

Item Type: Thesis (PhD)
Creators: Horváth Győző
Hungarian title label: Kisemlős populációk paramétereinek becslése és modellezése
Title of the thesis in foreign language: Estimating and modelling parameters of small mammal populations
Divisions: Doctoral School of Environmental Sciences
Discipline label: természettudományok > biológiai tudományok
Defence date label: 2009. May 21.
Item ID: 1258
Date Deposited: 2011. Nov. 18. 08:18
Last Modified: 2018. May. 17. 11:12
Depository no.: B 4573
Supervisor label:
Supervisor Supervisor scientific name label
Dr. Gallé László
egyetemi tanár, az MTA doktora, SZTE TTIK Ökológiai Tanszék
Reviewer label:
Reviewer name label Reviewer scientific name label
Dr. Demeter András
Dr. Csorba Gábor
PhD, Nature and Biodiversity Unit. Brüsszel
muzeológus, PhD Magyar Természettudományi Múzeum
President label:
President name label President scientific name label
Dr. Gausz János
tudományos tanácsadó, az MTA doktora, MTA SZBK Genetikai Intézet
Member label:
Member name label Member scientific name label
Dr. Györffy György
Dr. Margóczi Katalin
egyetemi docens, Ph.D., SZTE TTIK Ökológiai Tanszék
egyetemi docens, Ph.D., SZTE TTIK Ökológiai Tanszék
Defence/Citable status: Defended.

Actions (login required)

View Item View Item


Downloads per month over past year