Great crested newt (Triturus cristatus) as a diversity indicator species and evaluation of sampling methods
Table of contents
2 Introduction 3
3 Material and methods 5
3.1 Indicator species study 5
3.1.1 Field work 5
3.1.2 Analyses 6
3.2 Sampling variation study 6
3.2.1 Temporal variation 6
3.2.2 Interobserver variation 7
4 Results 7
4.1 Indicator species study 7
4.2 Sampling variation study 10
4.2.1 Temporal variation 10
4.2.2 Interobserver variation 12
5 Discussion 14
5.1 Indicator species study 14
5.2 Sampling variation study 15
5.2.1 Temporal variation 15
5.2.2 Interobserver variation 17
5.3 Conclusions 18
Indicator species are sometimes used to facilitate biodiversity monitoring, but the selection of indicator species requires careful evaluation of comprehensive data in order for conservation efforts to be successful. The Great crested newt (Triturus cristatus) has previously been proposed as a possible indicator species for aquatic macrophyte diversity. Six pairs of ponds in southeastern Sweden were examined for macroinvertebrate diversity. Each pair consisted of one pond with and one without T. cristatus, but otherwise as similar as possible. There were no significant differences between ponds with and without T. cristatus. However, differences within the pairs differed significantly between those pairs located in open areas, such as pastures or arable land, compared with pairs in forest environments. When comparing the two habitat types the open areas showed a pattern of high diversity in ponds with T. cristatus and low in ponds without whereas the forest habitat showed no clear pattern. Sampling error in suggested methods for T. cristatus monitoring was also evaluated, as were temporal and interobserver variation. Significant differences among observers were observed using visual sampling but not when dipnetting for T. cristatus larvae. The visual sampling method appeared to be biased in favor of males compared with other methods. Besides implications for methods, the information gathered could be used in poweranalyses to estimate sample size needed for effective monitoring of populations.
Keywords: amphibians, Caudata, interobserver variation, macroinvertebrate, temporal variation, visual observation
Amphibians are currently suffering declines at unprecedented rates worldwide and of the species described, 31% are listed as threatened (IUCN criteria CR, EN or VU). The number of threatened amphibian species has, during the past ten years (1996-2006), increased from 124 to 1811 and the number of critically endangered species has increased from 18 to 442. In comparison, the number of threatened species in any of the classes mammals, birds and reptiles have not increased even one tenth of this percentage. Among the amphibian orders, Caudata (newts and salamanders) has the highest percentage with 46.5% of all known species being threatened. Furthermore amphibian values are likely to be underestimated since approximately 24% of amphibian species described are classified as Data Deficient. The percentage of Data Deficient mammals, birds and reptiles are markedly lower. (IUCN Red List 2006). Analyses of earlier records show that the declining trends in amphibian populations started about five decades ago (Houlahan et al. 2000) and they have worsened during the last 25 years (Beebee & Griffiths 2005). The conclusion from these numbers of decreases and lack of data are that amphibians are in urgent need of monitoring and conservation efforts. This requires continued development of reliable methods for monitoring amphibian populations in order to determine conservation priorities as well as evaluating effects of conservation efforts (Buckley & Beebee 2004).
To facilitate monitoring indicator species/taxon are sometimes used e.g. as a measure of diversity among other taxa. But the selection of one or a few taxa as indicators for several others is far from unproblematic and requires a careful evaluation of the indicators based on comprehensive data. If not, there is a risk that conservation efforts are futile and resources wasted (Lawler et al. 2003, Lindenmayer&Fischer 2003, Grenyer et al. 2006). The steep decline in amphibian populations could indicate that they suffer a large impact from the environmental threats that has developed during the recent decades and could be among the most sensitive taxa to the effects of human activities. If this is the case they could possibly be used as indicators of unaltered ecosystems.
The great crested newt (Triturus cristatus Laurenti 1768), a caudate amphibian, has previously been proposed as an indicator species of aquatic macrophyte diversity (Gustafsson et al. 2006). Triturus cristatus adults migrate from their terrestrial habitat to breed in ponds during spring and early summer and the larvae develop in the aquatic habitat during the summer and thereafter become terrestrial at late summer or fall (Malmgren 2002). Since T. cristatus seems to display metapopulation dynamics, is sensitive to habitat fragmentation and is depending on connectivity between aquatic and terrestrial habitats (Joly et al. 2001) it could perhaps be a useful indicator for undisturbed pond landscapes and pond biodiversity.
In order to get reliable data for analysis when performing monitoring, it is crucial that the methods used are accurate (i.e. give a reasonable representation of the underlying “true values”, despite variation caused by varying conditions during sampling) and reliable (repeatable). Hence, it is important to know the reasons and magnitude of any distortions. Monitoring amphibian populations is complicated by low and variable probabilities of detection (Schmidt 2005). Except for small, well defined populations, the most practically feasible way of performing large scale monitoring of amphibians has been suggested to be by recording presence/absence instead of trying to measure changes in abundance in individual populations (Beebee & Griffiths 2005, Buckley & Beebee 2004). This is mainly due to the, compared with presence/absence sampling, more labour intensive methods of newt population size estimation for example by mark-recapture (Bailey et al. 2004). However, the reliability of such presence/absence data remains unknown.
One aim of this project was to investigate whether T. cristatus could be used as an indicator species for macroinvertebrate diversity, and another was to evaluate sampling variation when using the standardized methods proposed by the Swedish Environmental Protection Agency (Malmgren et al. 2005). Temporal and interobserver variations in T. cristatus abundance sampling will be described aiming for ways to improve the data quality of monitoring schemes and perform poweranalyses.
3 Material and methods
3.1 Indicator species study
3.1.1 Field work
In the indicator species study, six pairs of ponds that were pairwise similar with respect to size, depth, surrounding area and over-grown to the same extent were compared. Each pair consisted of one occupied pond where T. cristatus larvae were found and one empty pond with earlier record (Karlsson 2006) of adult absence and where no newt larvae were found. They were all situated in a landscape with regular occurences of newts and newt reproduction (Karlsson 2006, personal observations) so that absence in a sampled pond should not be primarily caused by isolation effects. Three of the pairs were located in open areas such as pastures or arable land (hereafter referred to as open areas) and three were located in forest dominated areas. Selection of which ponds to examine was based on records and maps of earlier observations of absence/presence and reproduction provided by the County administrative board in Östergötland (Karlsson 2006). In those records, some basic properties of the ponds were described and those combined with field observations were used to find ponds similar enough to form pairs. The ponds should not contain fish or crayfish. With the above mentioned criteria applied to records of 349 ponds in Östergötland surveyed for T. cristatus, six acceptable pairs were found. The chosen ponds were all within a range of approximately 45 km. The size range of the ponds was approximately from a couple of meters to slightly over twenty meters across.
Dipnetting of newt larvae was performed during August according to the Z-sweep method described by Malmgren et al. (2005). The dipnetting was performed by sweeping the net through the littoral vegetation at 2-5 dm above the bottom in a Z-like movement during approximately 3 seconds. This was repeated every 5 meters around the entire littoral zone or until larvae were found, and thereby verifying reproduction.
Invertebrate sampling was performed with a cylinder sampling of five randomly selected places in each pond at 2-4 dm depth. The samples were collected with a plastic cylinder of 33 cm in diameter that was pushed into the bottom sediment and thereby isolating a small body of water. This water and its contents was scooped out and filtered through a 0.5 mm mesh and the filtrate was used as one sample. Areas for sampling that were inaccessible or where the method did not work properly, e.g. due to roots or other hard structures on the bottom, were excluded during the selection.
Log-transformed abundance values were used in order to reduce the influence of the most abundant species. The macroinvertebrate assemblages in the ponds were described by a Principal Component Analysis (PCA). A comparison of the macroinvertebrates in ponds with and without T. cristatus larvae was done using partial Redundancy Analysis (pRDA), using the pairs as covariables (entered as a number of categorical dummy variables). In both cases, the software CANOCO 4.5 software (ter Braak and Smilauer 2002) was used. Differences within pairs were also tested in a paired t-test, with Simpsons diversity index and taxon count as response variables.
3.2 Sampling variation study
3.2.1 Temporal variation
In the study where the visual method described by Malmgren et al. (2005) was evaluated, two ponds were selected for an intensive study and, as described in the next section, for an interobserver variation study. One was a large pond with high abundance of T. cristatus and one was a smaller pond with lower abundance.
In the intensive study, ponds were sampled using visual observation approximately two times per week and on each occasion usually twice. Sampling was performed during approximately two hours after dark.
Visual observation was performed during May and the first half of June according to the standardized methods described by Malmgren et al. (2005). The observation was performed at night using a relatively powerful (10-20 W) headlight. The observer walked slowly around the water´s edge while looking for newts and stopped every 5 m to more carefully look during approximately 30 seconds. The whole pond was searched and four classes of individuals were noted: males, females, juveniles and adults that could not be sexually determined.
Pondwise standardized values of abundance were plotted as function of date and minutes after sunset when sampling was performed and lowess-lines (stiffness 0.6) were used to illustrate trends in data. The sex ratio during the season was plotted with 95% binomial confidence intervals (Sauro & Lewis 2005). Analysis of sample size needed at different standard deviations to detect differences among T. cristatus populations was calculated using Studysize 1.0.
3.2.2 Interobserver variation
In the interobserver variation study, five observers sampled the ponds during two nights each in May using the visual observation method. Furthermore, the same observers also dipnetted for larvae twice during August. Visual observation and dipnetting of newt larvae was performed according to the standardized methods described by Malmgren et al. (2005) and briefly in the previous sections.
Interobserver variation was analysed using variance component analysis in the STATISTICA software version 7.0 (StatSoft, Inc. 2004).
4.1 Indicator species study
The pRDA showed no significant differences (P=0.3203) in macroinvertebrate assemblages between ponds containing T. cristatus larvae and those that did not. PCA was used to further illustrate differences in species composition between ponds in the selected pairs, but no clear patterns within the pairs was found (Figure 1). A paired t-test performed on taxon count showed no significant differences within the pairs (P= 0.298, t(5)=1.1622), and the same test performed using Simpsons diversity index (1-D) values did not show differences either (P=0.881, t(5)=0.1569).
However, a comparison of pairs located in different areas did show some differences. When comparing Simpsons diversity index (1-D) the pairs in forest environments differed from the ponds located in open environments, such as pastures or arable land, which all had at least 59% higher diversity index in ponds where newt larvae were present than in the ones where they were absent (Figure 2). No significant difference in diversity index within pond pairs in forest versus open area was shown (P=0.0824, t(2.1)=3.1493) using a t-test, while differences in taxon count within pairs did show significant differences (P=0.035, t(4)=-3.1264). The ponds with presence in open landscapes had at least 78% higher taxon count and 59% higher diversity index than the absence ponds. There was no apparent pattern in variation in the pairs in forest environment.
Haliplus Haliplus sp
Figure 1. PCA of species composition in pairs of ponds with (filled circles) and without (empty circles) T. cristatus larvae. Ponds 1, 2 and 3 were located in open areas while 4, 5, and 6 were located in forest. There were 55 taxa in the analysis but only the 17 contributing most to variation are displayed (complete names in table). Eigenvalues of PC1 and PC2 were 0.275 and 0.213 respectively.
igure 2. Taxon count (2a) and Simpsons diversity index (1-D) (2b) for the six pairs of ponds. Pairs 1, 2 and 3 were located in open areas such as pastures or arable land while 4, 5 and 6 were located in forest environments
4.2 Sampling variation study
4.2.1 Temporal variation
Trends in seasonal variation appeared to consist of a small increase during the first part of the sampling period and a more pronounced, but still a slow, decrease towards the end (Figure 3a). Time of sampling during the night also showed a trend where peak values appeared to be reached around 90 min after sunset (Figure 3b).
The individuals of T. cristatus recorded changed from being clearly dominated by males during the first half to more equal, but still male-dominated counts in the second half of the sampling period (Figure 4). The equalization is caused by a steady decrease in male counts and a smaller increase in female counts in the second half.
Minutes after sunset
Figure 3: a) Seasonal variation in T. cristatus abundance. Standardized values from both ponds. b) Temporal variation of T. cristatus abundance. Standardized values from both ponds.
Figure 4. Seasonal variation in sex ratio (proportion of males) with 95% binomial confidence intervals.
4.2.2 Interobserver variation
The variance component analysis using standardized values from each pond showed a significant (P=0.0443)(F=6.88) difference between observers with 51% of the total variance being explained by variation between observers. The same analysis performed on values from larval sampling by dipnetting (only the large pond) showed no significant observer effect (Figure 5) (P=0.385)(F=1.29).
In the smaller pond with lower abundance, no T. cristatus were found in 60% of the samplings (Figure 5). This would result in a 36% probability of an inventory resulting in a “false zero”, when using the presently suggested methodology that consists of two samplings. If the criteria to consider T. cristatus as absent would be based on three instead of two negative samplings, the probability of an inventory resulting in a “false zero” would be 22%. In comparison, an observer with sampling experience found no T. cristatus in 28% of 18 samplings. This would result in an 8% probability of an inventory resulting in a “false zero” (2% if based on three negative samplings).
In the large pond, T. cristatus was found in every sampling. The number of individuals found varied between 10 and 39 (average 23.8, SD 10.75).
Figure 5. Interobserver variation in T. cristatus adults and juveniles (a and b) and in larvae (c). Observer 5 had previous sampling experience while observers 1-4 only had basic species information and method instructions. First sampling empty bars, second sampling filled bars.
5.1 Indicator species study
The results suggests that presence of T. cristatus larvae can not be used as an indicator of macroinvertebrate diversity under the circumstances in this study. However, the average taxon count for presence ponds was 46% higher than in the absence ponds. The surrounding area of the selected pairs could probably have contributed to a large part of variation in the data, since the results showed signifiant differences in taxon count within pairs from either open or forest environment. The eventual indicator species properties of T. cristatus could be restricted to only some of its habitats.
Studies on the use of indicator taxa have shown that they can be a useful tool in diversity assessments, but that indicator properties can be quite specific for certain taxa, spatial scales and geographic regions. Correlations between taxa may vary considerably between different geographic regions so that patterns of correlations between taxa in one area do not necessarily show the same pattern even in another similar area (Hess et al. 2006). Highest probability for a taxon to function as an indicator appears to be among taxonomically similar species (e.g. rare plants indicating plant richness) or taxa with functional dependence (butterflies as indicator for flowering plant richness: Flather et al. 1997). However, Lawler et al. (2003) found that threatened species from six different taxa performed better than the individual taxonomic groups (freshwater fish, birds, mammals, freshwater mussels, reptiles, amphibians) as indicators of overall taxonomic diversity. Regarding the spatial scale Grenyer et al. (2006) concludes that “even among terrestrial vertebrates, the extent to which rare and threatened species from one group can act as a surrogate for corresponding species in other groups is severely limited, especially at the finer scales most relevant to conservation”.
At a larger spatial scale, T. cristatus could possibly have indicator species properties other than eventual cross taxon biodiversity, for example as an indicator of an unfragmented landscape due to the species low dispersal abilities and dependence on metapopulation dynamics (Malmgren 2002, Joly et al. 2001). Hager (1998) proposes that fragmentation-sensitive species of amphibians and reptiles could be used as this type of indicator due to nested patterns of amphibian occurences found on islands. Hence, stable and functioning metapopulations of T. cristatus could possibly have value as an indicator of an unfragmented landscape and connectivity between ponds suitable for amphibians. Lambeck (1997) suggests, in accordance with this, that when selecting areas for reserves a suite of species should be selected which represents the most sensitive to one or more of the different threats that the reserve aims to protect from. The most sensitive species will thereby set the acceptable level of a certain stressor. The chosen species could e.g. be area-, connectivity- or resource demanding. Pearman (1997) and Krishnamurthy (2003) found that some amphibian taxa were clearly more affected by habitat fragmentation and degradation caused by human activities than other. Triturus cristatus eventual usefulness as a connectivity indicator requires further investigation.
Criticism against the use of indicator species rightfully claims that not all threats and their effects will be detected when using one or few species as indicators, but an advantage in focusing on the presumably most sensitive species to a known threat is that the eventual negative impacts will likely be detected earlier than when focusing on entire communities (Faria et al. 2006). Furthermore, the single/few species monitoring is likely less work-demanding than ecosystem/community monitoring and therefore requires less resources enabling monitoring that is spatially or temporally more comprehensive.
Further testing with a large number of pairs from increased sample areas including various kinds of habitats could perhaps shed more light on the possible indicator species properties for T. cristatus in different habitats, spatial scales and taxa.
5.2 Sampling variation study
5.2.1 Temporal variation
Triturus cristatus peaked at the early parts of May coinciding with the reproductive activity (Malmgren 2002). After the winter T. cristatus is depending on a temperature rise to 0-5°C and rainfall for initiation of migration to reproduction ponds (Malmgren et al. 2005).
The period of highest T. cristatus activity (above average) appears to be reached around 90 minutes after sunset (Figure 4) and thereafter stay at rather high levels during the sampled period, indicating optimal time of sampling to be from around 90 minutes to at least 210 minutes after sunset
The male-dominated sex ratio of observed T. cristatus could be caused by different probabilities of detection of males and females since previous studies, using other than visual methods, have shown that the only skew in sex ratio for T. cristatus generally is towards a higher proportion of females in the later parts of the breeding season (Arntzen 2002). A large part of the observed difference could possibly be caused by the easily spotted male bright white stripe along the tail. This stripe allows easy spotting as well as identification of males but is not seen from all angles causing the ratio of sexually unidentified newts to likely be skewed toward higher proportion of females. This could in turn cause a bias in numbers of observed newts of each sex since the sexually unidentified individuals were excluded from the sex-ratio calculations. A personal observation was that T. cristatus (primarily males) often appeared to aggregate in open areas while observations from more densely vegetated areas gave more equal counts of males and females. These aggregations are likely a display of lekking behaviour (Hedlund 1989) and if males gather in more open areas it could increase their probability of detection and cause a further skew in the sex ratio.
The visual observation method (Malmgren et al. 2005) as a measure of population size is likely to be a relatively blunt instrument in estimating population size compared to more resource demanding techniques such as mark/recapture (Bailey et al. 2004, Schmidt 2004, Flint & Harris 2005). However, visual encounter methods have, for other caudates, shown to give a valid index of relative population sizes (Flint & Harris 2005). It could probably have a value as a complement to presence/absence recordings considering the relatively low increase in workload to count individuals in ponds sampled. Figure 6 illustrates how the minimal detectable difference varies with SD and sample size using a paired t-test. Using the estimate of SD of the abundance of T. cristatus from the large pond, (SD=10.8 when using data from five observers) we can estimate the number of ponds needed to detect an average change in abundance estimated at two points in time. Although number of ponds is likely to be seen as a minimum value, since the values are based on a pond with relatively high abundance and large in size (smaller ponds will likely have a higher degree of sampling error causing variation; Marsh 2001). Caution is generally needed when drawing conclusions from amphibian abundance data, due to often high seasonal fluctuations in population Coefficient of Variation (CV=SD/(Average number of individuals); Marsh 2001). The CV for the ponds in the intensive study were 0.407 and 0.947 for the large and small pond, respectively.
The variation in abundance data can also be used for poweranalyses for longer series. For example four sampling series conducted every other year in the large pond would detect, at the 5% level, a 10% decrease in abundance nine times out of ten (estimated using MONITOR 7.0; Gibbs 1995)
Longer time series from several ponds, with as little inter-pond correlation as possible, might be necessary to differentiate variation caused by natural fluctuations from that of antropogenic disturbances (Pechmann et al.1991).
Figure 6. Estimation of number of ponds needed to detect population changes (Significance level=0.05, Power=0.8, H0: Mean of difference =0)
5.2.2 Interobserver variation
The significant differences between observers could probably in part be explained by one observer having field experience of the method used when this sampling was performed compared with observers with no previous experience. The difference between observers appeared most pronounced with regard to be able to distinguish juvenile T. cristatus from adult T. vulgaris at a distance, and this will likely demand more practice and experience than sampling adults. The most pronounced difference at a distance is the heavier built T. cristatus juveniles with a relatively larger head, but some practice is needed to easily see the difference. There are also differences in markings but they are less obvious when seen from above in suboptimal light and visibility conditions as often is the case.
The probability of two consecutive samplings not detecting T. cristatus presence in the low abundance (small) pond by inexperienced observers compared with a somewhat experienced observer was in this case four times higher (0.36 vs 0.08). This is presumably to a large part caused by the above-mentioned difficulty in distinguishing T. cristatus juveniles from T. vulgaris adults since the probability of the somewhat experienced observer getting a false zero increases to 0.20 if excluding juveniles from the data. This could indicate that inexperienced observers need to sample the same pond more than twice in order to be able to determine absence with the same level of confidence as an experienced observer. Increasing the number of samplings to three or four would decrease the probability to 0.22 and 0.13 respectively. Alternatively the samplings could be performed with a couple of weeks separation, thereby allowing the observer to gain experience between samplings.
The reason that sampling of larvae showed no significant inter-observer differences is likely to be a combination of the fact that the dipnetting method does not require ability to quickly determine species at a distance, in contrast to the visual observation, and a relatively small number of samplings. However, the larval sampling gave much lower abundance values and is therefore not preferable compared with the visual method in determining presence/absence due to the higher risk of “false zeros” at low abundances.
Triturus cristatus does not appear to be useful as an indicator of macroinvertebrate diversity under the circumstances in this study, but further research in different spatial scales, habitat types and properties being indicated should be undertaken before drawing final conclusions. The monitoring methods of T. cristatus used gave a relatively low amount of incorrect absence/presence assessments and appear to be accurate enough to delineate expected trends in temporal variation. Interobserver variation was evident, likely in large part due to different amounts of sampling experience.
I thank my supervisors Per Milberg and Karl-Olof Bergman for their support. I would also like to thank Tommy Karlsson at the County Administrative Board in Östergötland and Jan Malmgren for their support with species and methods information, and Anders Göthberg at Linköping University for his help with invertebrate sampling and taxon determination.
Arntzen JW (2002) Seasonal variation in sex ratio and asyncronous presence at ponds of male and female Triturus newts. Journal of Herpetology 36, 30-35
Bailey LL, Simons TR, Pollock KH (2004) Estimating site occupancy and species detection probability parameters for terrestrial salamanders. Ecological Applications 14, 692-702
Beebee TJC, Griffiths RA (2005) The amphibian decline crisis: a watershed for conservation biology? Biological Conservation 125, 271-285
Buckley J, Beebee TJC (2004) Monitoring the conservation status of an endangered amphibian: the natterjack toad Bufo calamita in Britain. Animal Conservation 7, 221-228
Faria MS, Ré A, Malcato J, Silva PCLD, Pestana J, Agra AR, Nogueira AJA, Soares AMVM (2006) Biological and functional responses of in situ bioassays with Chironomus riparius larvae to assess river water quality and contamination. Science of the Total Environment 371, 125-137
Flather CH, Wilson KR, Dean DJ, McComb WC (1997) Identifying gaps in conservation networks: of indicators and uncertainty in geographic-based analyses. Ecological Applications 7, 531-542
Flint WD, Harris RN (2005) The efficiacy of visual encounter surveys for population monitoring of Plethodon punctatus (Caudata: Plethodontidae) Journal of Herpetology 39, 578-584
Gibbs JP (1995) MONITOR 7.0
http://www.mbr-pwrc.usgs.gov/software/monitor.html (9 mars 2007)
Grenyer R, Orme CDL, Jackson SF, Thomas GH, Davies RG, Davies TJ, Jones KE, Olson VA, Ridgely RS, Rasmussen PC, Ding TS, Bennet PM, Blackburn TM, Gaston KJ, Gittleman JL, Owens IPF (2006) Global distribution and conservation of rare and threatened vertebrates. Nature 444, 93-96
Gustafsson DH, Petterson CJ, Malmgren JC (2006) Great crested newts (Triturus cristatus) as indicators of aquatic plant diversity. Herpetological Journal 16, 347-352
Hager HA (1998) Area-sensitivity of reptiles and amphibians: are there indicator species for habitat fragmentation? Ecoscience 5, 139-147
Hedlund L, Robertson JGM (1989) Lekking behaviour in crested newts, Triturus cristatus. Ethology 80, 111-119
Hess GA, Bartel RA, Leidner AK, Rosenfeld KM, Rubino MJ, Snider SB, Ricketts TH (2006) Effectiveness of biodiversity indicators varies with
extent, grain, and region. Biological Conservation 132, 448-457
Houlahan JE, Findlay CS, Schmidt BR, Meyer AH, Kuzmin SL (2000) Quantitative evidence for global amphibian population declines. Nature 404, 752-755
International Union for Conservation of Nature and Natural Resources IUCN Red List of endangered species. http://www.iucnredlist.org/info/stats (14 mars 2007)
Joly P, Miaud C, Lehmann A, Grolet O (2001) Habitat matrix effects on pond occupancy in newts. Conservation Biology 15, 239-248
Karlsson T (2006) Större vattensalamander (Triturus cristatus) i Östergötland: Sammanställning av inventeringar 1994-2005 och övriga fynd i Östergötlands län. Länsstyrelsen Östergötland, rapport 2006:4.
Lawler JJ, White D, Sifneos JC, Master LL (2003) Rare species and the use of indicator groups for conservation planning. Conservation Biology 17, 875-882
Lindenmayer DB, Fischer J (2003) Sound science or social hook-a response to Brooker’s application of the focal species approach. Landscape and Urban Planning 62, 149-158
Malmgren J, Gustafsson D, Pettersson CJ, Grandin U, Rygne H (2005) Inventering och övervakning av större vattensalamander. Naturvårdsverket
Malmgren J (2002) Faktablad: Triturus cristatus – större vattensalamander. ArtDatabanken.
Marsh DM (2001) Fluctuations in amphibian populations: a meta-analysis. Biological Conservation 101, 327-335
Pechmann JHK, Scott DE, Semlitsch RD, Caldwell JP, Vitt LJ, Gibbons JW (1991) Declining amphibian populations: the problem of separating human impacts from natural fluctuations. Science 253, 892-895
Schmidt BR (2005) Monitoring the distribution of pond breeding amphibians when species are detected imperfecly. Aquatic Conservation: Marine and Freshwater Ecosystems 15, 681-692
Schmidt BR (2004) Declining amphibian populations: the pitfalls in count data in the study of diversity, distributions, dynamics, and demography. Herpetological Journal 14, 167-174
StatSoft, Inc. (2004) STATISTICA (data analysis software system), version 7, www.statsoft.com
Sauro J, Lewis JR (2005) Estimating completion rates from small samples using binomial confidence intervals: Comparisons and recommendations. In Proceedings of the Human Factors and Ergonomics Society 49th Annual Meeting (pp. 2100-2104). Santa Monica, CA: Human Factors and Ergonomics Society. http://www.measuringusability.com/wald.htm
(14 mars 2007)
Studysize 1.0 CreoStat HB. Enbarsv.11. 42655 V.Frolunda. Sweden.
ter Braak CJF Smilauer P (2002) Canoco reference manual and user’s guide to Canoco for windows: software for Canonical Community Ordination (version 4). Microcomputer Power, Ithaca, NY.