The first dataset compiled for the study includes reported catch records of the focal species for which the recreational angling magazines “Rybářství”, “Rybář”, “Český Rybář”, “Sportovní Rybářství”, “Kajman” and the recreational angler´s website “mrk.cz” were examined for evidence of record fish of the species crucian carp, gibel carp, common bream and roach from the years 1973–2022, resulting in a dataset spanning 50 years. The submitted photos, together with the total length (TL, in cm), weight (kg; W) and angling area code, were first reviewed by a magazine editor-in-chief or an expert (an experienced angler and often an authority from the Czech or Moravian Anglers Union, or from the anglers’ community on the mrk.cz website) before being officially admitted to the competition. Selected photos appeared directly in the magazines, while the rest of the record catches were not selected for publication in the magazines and were included in the annual summarised statistics of record catches. Both datasets were extracted in a standardised manner, focusing on the presence of TL, W, species, water type (lentic, lotic) and numerical angling area identifier (code unique to a particular water body or river section, if available in the record). In addition, due to the similarity between crucian carp and gibel carp, the authors’ team reviewed all available photos to assess the reliability of the dataset and changed the category from crucian carp to gibel carp (or vice versa) where appropriate. We also extracted the central GPS position of the angling site and the total area in ha and assigned the angling ground to three existing watersheds in Czechia (Elbe, Danube and Odra; Suppl. material 1).

The angling sites are part of the Czech and Moravian Anglers Unions, which are large organisations with around 330,000 members and whose angling grounds cover the catchment areas of the Elbe, Danube (Morava) and Odra Rivers in Czechia. There are 76,000 kilometres of watercourses and 107 reservoirs in Czechia. The area of waters totals 42,000 hectares, both in lotic and lentic ecosystems and is accessible for recreational angling with a licence.

To estimate whether the number of record-sized individuals of a given species follows any pattern in the time series, the data were tested with the funtimes package (Lyubchich and Gel 2023) using the local regression-based WAVK test method (Wang et al. 2008; Lyubchich et al. 2013) within the R software (R Core Team 2023). The Sieve bootstrap enhancement to test for a trend (monotonic or non-monotonic) was used with the WAVK function for each species separately (Lyubchich et al. 2013).

In addition, the records of native crucian carp and invasive gibel carp were analysed to determine whether they differ in their maximum recorded length and weight. The crucian carp and gibel carp datasets were tested for normality assumptions using the Shapiro-Wilk normality test. As the data were not normally distributed, the Mann-Whitney U-test was used for all data regardless of the year of capture.

Generalised additive models (GAM) were used to assess trends in fish size during the study period (Wood 2017). To test whether the trend in maximum length and weight changed over the years, two GAMs were created with all species in the first stage, with length and weight as response variables and species, year and angling ground size as explanatory variables. To check the validity of k-value, the gam.check function was used (Augustin et al. 2012; Wood 2017). In addition, a total of eight GAMs were created (for each species separately) with length and weight as response variables and year as an independent variable to assess species-specific trends over the years. The effect of the variable year was modelled using cubic regression splines (bs = “cr”). General additive models were created using the mgcv package (Wood 2001, 2017).

Generalised additive models were used to plot the points of records of all four species for each decade, using the number of reported catches in each period as the response variable and their GPS locations as the explanatory variable, by creating contour plots (Wood 2017). Each line (or contour) represents the number of reported catches in that area within the country. The proximity of the lines indicates the steepness of the gradient. The model check was performed using the gam.check function (Augustin et al. 2012; Wood 2017). The graphical visualisation of the data was created using the ggplot2 and ggmap packages (Kahle and Wickham 2013; Wickham 2016).

The test for any trend on all four species indicated that all species contain a significant trend in their data (WAVK test: crucian carp p < 0.001; gibel carp p < 0.001; common bream < 0.001; roach < 0.05). Trends in number of record crucian carp reported by recreational anglers declined sharply after 2005 and, for these data and the best fit, we used a model with moving window (MW) 7 and a polynomial fit of degree 11 (WAVK test = 34.098, p < 0.001). Reports of invasive gibel carp first appeared in 1985 and were comparable in number to native crucian carp between 1985 and 2000. Since then, reports of invasive gibel carp have become very dominant in terms of record sizes (Fig. 1A). The best fit for the gibel carp was a linear increasing trend (WAVK test = 152.51, MW = 7, p < 0.001). In comparison to this trend, the record size reports for common bream and roach did not show a strong decreasing or increasing trend (Fig. 1B, C). The best fits were a polynomial fit of degree 3 in common bream (WAVK test = 35.996, MW = 9, p < 0.001) and a linear trend in roach (WAVK test = 40.395, MW = 9, p = 0.021).

Figure 1.

The number of record catches extracted from angling magazines and websites between 1973 and 2022 for A the native crucian carp (Carassius carassius), invasive gibel carp (C. gibelio) and reference fish species B common bream (Abramis brama) and C roach (Rutilus rutilus). The period captures the invasion phase of gibel carp from early invasion phase to fully established, as well as the current impact on the native and critically endangered crucian carp in Czechia.

The reported native crucian carp were, on average, smaller than invasive gibel carp in both length (39.8 ± 4.3 cm vs. 44.5 ± 4.0 cm, W = 6167.0, p < 0.001; Fig. 2A) and weight (1.46 ± 0.43 kg vs. 1.82 ± 0.50 kg, W = 8282.5, p < 0.001; Fig. 2B). The general additive model for maximum recorded length (M1) and weight (M2) differed significantly between fish species, with the exception of roach length (Table 1), with a positive estimate for gibel carp and common bream and a negative estimate for roach in terms of weight and, further, with a positive effect of angling ground size (M1: t = 2.36, p = 0.018; M2: t = 2.34, p = 0.020) and a positive effect on weight in the Elbe catchment (M2: t = 2.28, p = 0.023). The effect of year was significant for both models (M1 F = 4.844, p < 0.001; M2: F = 3.936, p < 0.001) and the total explained deviation of the model was 83.4% and 71.8% for M1 and M2, respectively. The trend for the recorded maximum lengths was unimodal for native crucian carp with maximum values around the year 2000 (GAM: F = 15.12, p < 0.001, 33.4% deviance explained), while it gradually increased for invasive gibel carp (GAM: F = 18.84, p < 0.001, 24.1%), was relatively stable for common bream (GAM: F = 0.009, p > 0.05, 0.0%) and showed a fluctuating trend with the maximum around 1990 in roach (GAM: F = 3.841, p < 0.001, 14.7%; Fig. 3). Similarly, the maximum weight was recorded around the year 2000 for crucian carp (GAM: F = 21.93, p < 0.001, 44.7%), while the maximum recorded weight increased gradually in gibel carp (GAM: F = 3.726, p = 0.002, 8.5%) and the fit was linear in common bream (GAM: F = 2.335, p = 0.014, 7.6%) and fluctuating in roach (GAM: F = 10.19, p < 0.001, 16.7%; Fig. 4).

Figure 2.

The comparison of reported record catches size in the 50-year dataset. The maximum attainable size of the invasive gibel carp (Carassius gibelio) significantly exceeds the maximum size of the native crucian carp (C. carassius) in Czechia in both A total length and B weight. The boxplot boundaries represent upper and lower quartiles; the thick lines represent medians and the whiskers represent 1.5 times the interquartile range. Violin plots represent kernel density distribution.

Figure 3.

The GAM trend-line of record catches size (total length) in the 50-year dataset. The GAM fit has been computed with the stat_smooth function. The data suggest: A unimodal response in the native crucian carp (Carassius carassius) with time B increase maximum attainable size in the invasive gibel carp (C. gibelio) C stabilised size limits in common bream (Abramis brama) and D maximum in 1990 in roach (Rutilus rutilus).

Figure 4.

The GAM trend-line of record catches size (weight) in the 50-year dataset. The fit has been computed with the smooth function with a value of k = 1 to avoid overfitting. The data suggest A unimodal response in the native crucian carp (Carassius carassius) with time B moderate increase in maximum attainable size in the invasive gibel carp (C. gibelio) C relatively stabilised size limits in common bream (Abramis brama) and D fluctuating trend in roach (Rutilus rutilus).

The GAM contour plots showed the concentration of large-sized fish mainly in the area of the lowland rivers and the interconnected oxbows. While common bream and gibel carp were almost ubiquitous, record-sized crucian carp and roach showed a more aggregated distribution with few hotspots in Czechia (Fig. 5).

Figure 5.

Contour plots based on GAMs for A native crucian carp (Carassius carassius) B invasive gibel carp (C. gibelio) C common bream (Abramis brama) and D roach (Rutilus rutilus) for all study period, where the number of reported record catches in a given area was used as the response variable and their GPS locations as the explanatory variable. Each line (or contour) represents the reported catches in that region within the country. The proximity of the lines indicates the steepness of the gradient. Colours show the abundance of records with red indicating the highest abundance.

Both native crucian carp and invasive gibel carp were found more frequently in the lentic waters, which is consistent with their ecology (Holopainen et al. 1997; Tarkan et al. 2023). The size of the angling water and the catchment area of the Elbe had a slight positive influence on fish size in the general model for all species. This result may need to be tested in more detail using a larger dataset, as not much scientific literature has been published on angling for fish of record size and such result may apply only to some species.

While the records of both reference species showed a relatively stable trend around the same average value, both native crucian carp and invasive gibel carp showed GAM trend-lines indicating changes in maximum size. For crucian carp, the unimodal response with a decline in recorded maximum size in recent years suggests either growth limitations or possible confusion with invasive gibel carp or hybrids between crucian carp and gibel carp around 2000. Both options are possible, as the growth restriction may be caused either by increased interspecific competition due to the invasion of the gibel carp (Auwerx et al. 2021; Tapkir et al. 2022) or by the negative effects of increasing average temperature on fish growth (Emmrich et al. 2014). While the prediction of crucian carp via growth parameters suggests that growth increases with temperature (Tarkan et al. 2016), the record-size crucian carp seems to benefit from rather low temperatures and the best lakes are located in northern latitudes and with the presence of piscivorous fish (Rolfe 2010; Vinterstare et al. 2023).

Hybridisation is also a likely explanation, as both species form hybrids under certain circumstances (Papoušek et al. 2008; Knytl et al. 2018) and hybrids also form between the crucian carp and the goldfish (Smartt 2007). Despite the emphasis on hybridisation in Carassius studies (Papoušek et al. 2008; Wouters et al. 2012; Knytl et al. 2018, 2022), the results of strong invasive gibel carp suggest that this is competition between native crucian carp and probably the main reason for the decline in crucian carp populations (Tapkir et al. 2022, 2023).

The gibel carp was accidentally introduced to Hungary together with the commercial Asian fish species (Hypophthalmichthys molitrix, H. nobilis, Ctenopharyngodon idella) in the 1950s (Tóth 1976; Holčík 1980). The westward invasion of the gibel carp started in the Danube catchment, and commercial catches of Carassius genus increased from 3% to 15% in the period from 1958 to 1976 in the Danube (Tóth 1976; Holčík 1980) and the first records in Czechia were in the Morava River (Danube watershed) in 1975–1976 (Halačka et al. 2003). Due to inter-basin aquaculture transfers, the first reports of gibel carp in the Elbe River Basin were recorded as early as 1980 (Kubečka 1989) and soon became the dominant taxa within the genus Carassius (Halačka et al. 2003; Lusk et al. 2010). The first record in this dataset also comes from the Danube catchment. However, this study focused on fish of record size, so it is likely that several years will pass between the invasion and the record catches.

There are currently three invasive species of the genus Carassius living in Czechia, the goldfish (including feral form), C. langsdorfii in addition to already mentioned gibel carp (Kalous et al. 2007, 2013; Rylková et al. 2013). These species can hybridise with each other and form viable populations of hybrid origin (Keszte et al. 2021) and are, therefore, considered a species complex in some studies (Rylková et al. 2013; Knytl et al. 2022). All of them are of similar appearance (Hensel 1971), the largest individuals can reach a length of more than 40 cm and, therefore, all can be included in angling statistics under the name of invasive gibel carp. Although it is widely believed that the gibel carp is the most widespread invasive Carassius species in Central Europe and the other two species are of lesser importance in terms of competition with native crucian carp (Lusk et al. 2010; Musil et al. 2010; Fedorčák et al. 2023), it might be interesting to investigate the relative proportion of these invasive species in European mainland.

The passion for enhancing native crucian carp populations was first brought to public attention in England, where Peter Rolfe launched his attempt as a pilot angler to reintroduce the species to Norfolk waters and promote the existence of exceptionally large fish in ponds (Copp and Sayer 2010; Rolfe 2010). The approach has been underpinned by the scientific literature and the presence of European pike (Esox lucius) has been embedded in the management of larger ponds to encourage the presence of the deep-bodied phenotype that can reach exceptional sizes that are attractive to recreational anglers (Brönmark and Miner 1992; Brönmark et al. 1995; Pettersson and Bronmark 1997; Rolfe 2010). This concept involves easing the intraspecific competition through predation (i.e. thinning out the carp population in the water body) and the simultaneous production of crucian carp in small ponds and stocking in larger water bodies.

Although the conservation and fisheries management described above is artificial, it has its roots in the life-history strategy of the crucian carp. In the floodplain rivers, the population has a metapopulation structure with a rare deep-bodied phenotype in multi-species community of large water bodies or lowland rivers (Brönmark and Miner 1992; Brönmark et al. 1995; Holopainen et al. 1997). These individuals can colonise pools in the floodplain and lay the foundation to the shallow-bodied phenotype that is formed in a single-species community without the presence of piscivores due to the frequent occurrence of anoxia (Blažka 1958; Holopainen and Hyvärinen 1985; Piironen and Holopainen 1986). These pools are then source populations and provide a surplus of fish in the main river channels during floods, where they can produce deep-bodied and potentially large-sized fish. From the perspective of Czechia, it appears that the deep-bodied phenotype disappeared faster than the shallow-bodied phenotype, as citizen-science projects resulted in finding few tens of sites (Šmejkal et al. 2021). Water Framework Directive monitoring in the Vltava watershed has not detected crucian carp in the last 15 years, while gibel carp are abundant (Bartoň and Šmejkal 2022), so we can assume that the trend presented in the catch records is reliable.

Culturomics in conservation has been shown to be an effective means of raising public understanding, framing conservation issues and engaging people in timely environmental monitoring (Ladle et al. 2016). Recreational angling with catch-and-release ethics for each species is a conservation strategy and these contests have been shown to be effective in raising awareness of species conservation and getting people’s attention (Cooke and Suski 2005). The catch data of mahseer (Tor spp.) from the Cauvery River in India is a good example of how organised angling can be used as a tool to monitor species conservation, as it effectively contributes to species conservation and fisheries management measures (Pinder et al. 2015). The appreciation of environmental goods and services is often a part of societal culture and digitised or documented information, such as angler logbooks, helps to understand the behaviour of recreational anglers towards a species (Hutt et al. 2013) and can be used for conservation and management measures via culturomics. Angler’s ecological knowledge, which largely depends on the size and frequency of their catches (van den Heuvel and Rönnbäck 2023), can be an effective tool for conservation alongside the cultural influences on their interpretations of environmental change (Thornton and Scheer 2012). There are approximately 226 million recreational anglers worldwide (World Bank 2002; Arlinghaus et al. 2015, 2019) and ~ 3% of the total population of Czechia are registered in angling unions (Boukal et al. 2012), so this angling information, if systematically retrieved and analysed, can provide a good source of information for conservation efforts.

The authors have declared that no competing interests exist.

This study did not involve handling of animals or personal information.

The collection of data resulting in this manuscript has been supported by the programme of Regional Cooperation of the Czech Academy of Sciences (R200962201) and the Research Programme Strategy AV21 Water for life for valuable support.

Marek Šmejkal: Conceptualisation, Investigation, Data curation, Formal analysis, Writing Original draft, Writing, Review & Editing, Visualisation. Kiran Thomas: Formal analysis, Visualisation, Writing, Review & Editing. Vladimír Kořen: Conceptualisation, Writing, Review & Editing. Jan Kubečka: Conceptualisation, Data curation, Writing, Review & Editing.

Marek Šmejkal https://orcid.org/0000-0002-7887-6411

Data will be shared by authors upon reasonable request.