Spatial and Temporal Trends in the Invasion Dynamics of the Ring-Necked Parakeet (Psittacula krameri) in the Urban Complex of Thessaloniki, Greece

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

It is interesting, well prepared manuscipt,  concerning an invasive species of parrot, its population abundance and habitat requirements in Thessaloniki. Clarifications are needed mainly  in the methodology. 

Comments for author File: Comments.pdf

Author Response

Response to Reviewer 1

The authors would like to thank Reviewer 1 for their time and effort invested in reviewing our manuscript. Their comments were very constructive and helped us improve our work.

Below, we provide a point-by-point response to the reviewer’s comments and suggestions.

All changes/additions are highlighted in the revised version of our manuscript.

Comment 1

Line 3: Suggestion to remove “the urban complex of” from the manuscript title.

Response: While we greatly appreciate the reviewer’s recommendation, we have decided not to make any changes in the title. This decision is mainly based on the fact that including only “Thessaloniki” in our title could be misleading. One could interpret that our study refers to the Municipality (city) of Thessaloniki (which occupies approximately 19.3 km²) or the Regional Unit of Thessaloniki which includes urban, sub-urban, rural and semi-rural areas (which occupies approximately 3 683 km²). Conversely, the term “urban complex” of Thessaloniki describes the city of Thessaloniki, plus the surrounding municipalities that have merged into one continuous urban fabric and which is described in detail in Line 107: “Thessaloniki is located in northern Greece, within the region of Central Macedonia, and is the country’s second-largest urban center after Athens. The city forms part of the Metropolitan Unit of Thessaloniki (MUTH), an administrative subdivision of the region that includes 13 municipalities [42]. The main urban complex consists of seven contiguous municipalities; Thessaloniki, Ampelokipoi-Menemeni, Kalamaria, Kordelio-Evosmos, Neapoli-Sykies, Pavlos Melas and Pylaia-Chortiatis, which together make up the city’s continuous built-up area (Figure 1).” Based on the above, we did not make any changes to the original title of the revised manuscript.

Comment 2

Line 25: This sentence implies that the parrot requires protective measures.

Response: Agreed. Has been modified according to the reviewer’s comment.

Line 24: “Understanding their spatial and temporal dynamics is essential for effective management planning and implementation”

Comment 3

Line 28: In which years?

Response: Agreed. Has been added according to the reviewer’s suggestion. Line 27: “During 2024 and 2025, we conducted repeated transect surveys across 99 1 km² grid squares within the urban complex of Thessaloniki to assess environmental factors influencing occupancy and abundance, and to estimate RNP population trends.”

 

Comment 4

Line 68: Remove reference number 7.

Response:  Agreed, reference has been removed.

Comment 5

Line 69: I am not convinced that citing so many publications is justified.

Response: Agreed. References previously numbered [7], [9], [18] and [19] have been removed due to their more general context regarding the statement. 

Comment 6

Line 72: I am not sure, that citing so many publications is justified.

Response: Agreed, reference previously numbered [29] has been removed. All other cited references support the statement.

Comment 7

Line 74: I am not sure, that citing so many publications is justified.

Response: Agreed, reference previously numbered [33] has been removed. All other cited references support the statement.

Comment 8

Line 95: Research hypothesis is missing.

Response: We thank the reviewer for this important observation. We have revised the Introduction to explicitly state our research hypotheses, now included in Line 96: “In this context, our study offers the first detailed analysis of the spatial and temporal trends of RNP expansion in Greece, focusing on the urban complex of Thessaloniki. Based on previous studies regarding the species, we hypothesized that (i) RNP occupancy and abundance would be positively associated with urbanization intensity and greater availability of nesting and foraging sites, and (ii) the population would show an increasing trend between survey years, reflecting the species’ ongoing spread and expansion. Our study could provide insights into the mechanisms that facilitate RNP establishment and persistence in Mediterranean urban environments and could serve as a baseline for long-term monitoring of the species’ spread and population size, and for assessing the outcomes of management and control measures.”

Comment 9

Line 124: One reference will be enough.

Response: Agreed, we kept the first reference as it was the newest. All other references have been removed.

 

Comment 10

Line 144: Too many references.

Response: Agreed, references previously numbered [20] and [55] have been removed.

Comment 11

Table 1: Please define the parameters.

Response: Thank you for pointing out this issue. The parameters included in the dynamic occupancy model have now been clearly defined in the Data analysis section of the revised manuscript, Line 172: “RNP detection and non-detection data were analyzed using multi-season dynamic occupancy models, wherein initial occupancy (ψ₁; i.e. the probability that a site was occupied during the first year of surveys), colonization (γ; i.e. the probability that an unoccupied site becomes occupied between survey years), extinction (ε; i.e. the probability that an occupied site becomes unoccupied between survey years), and detectability (p; i.e. the probability of detecting the species during a survey, conditional to its presence) are jointly estimated to describe state transitions across the surveyed area.”

Comment 12

Table 1: Please define GHII

Response: Agreed. GHII has been defined within Table 1. Line 168: “*GHII is a global index that measures cumulative human pressure based on population density, land use, infrastructure, and accessibility”

Comment 13

Line 155: How is that possible, since the counts were only taken in two seasons?

Response: Thank you for raising this important issue. I believe the confusion rises from the term “multi-season”, which in the occupancy literature specifically refers to any study with at least two primary periods (in our case, years 2024 and 2025) where the state of occupancy can shift (occupied/unoccupied). A two-year-primary design is the fundamental, minimal design allowing the implementation of this type of analysis. Our study has two primary periods; 2024 (Season 1) and 2025 (Season 2). Within each primary period, we have two secondary sampling occasions (repeated visits), which are used to account for detection probability. The primary strength of using this model, even with two years, is that it correctly accounts for detection error while separating the process of occupancy change from the observation of it.

Comment 14

Lines 158-160: In that case, why was it introduced into the model?

Response: Thank you for this critical question. We agree that testing covariates for extinction (ε) is often a key part of dynamic occupancy analysis. However, in our study no sites that were occupied during the first year of surveys were detected as unoccupied in the second year. If we tried to fit a covariate to ε, model estimation would fail since optimization algorithms cannot estimate coefficients for a process not observed in the data. Therefore, we constrained ε(.) to be constant across all models. This is a standard and necessary approach in such cases, as it reflects the empirical reality of our data over the 2-year study period, and allows the model to focus on estimating the parameters for which we have information.  Moreover, it is not statistically possible to exclude the extinction parameter and still fit a dynamic occupancy model, since all parameters are jointly estimated. We have now clarified this critical point in line 181: “Finally, because no instances of local extinction were detected during the second survey year (i.e., all sites occupied in 2024 remained occupied in 2025), extinction probability (ε) was held constant throughout the modeling process. This constraint reflects the observed biological pattern during the study period and ensures model estimability.” In addition, in line 172 it is now clearly stated that all model parameters are jointly estimated: “RNP detection and non-detection data were analyzed using multi-season dynamic occupancy models, wherein initial occupancy (ψ₁; i.e. the probability that a site was occupied during the first year of surveys), colonization (γ; i.e. the probability that an unoccupied site becomes occupied between survey years), extinction (ε; i.e. the probability that an occupied site becomes unoccupied between survey years), and detectability (p; i.e the probability of detecting the species during a survey, conditional to its presence) are jointly estimated to describe state transitions across the surveyed area.”

Comment 15

Lines 199-200: This is methods not results.

Response: Agreed, has been moved accordingly. Line 128: “Data describing the presence and abundance of RNP were obtained from field surveys conducted during what is believed to be the species’ breeding period. The study area was divided into 1 km2 grid squares [46], from which 103 were selected for potential surveying (Figure 1). The survey was designed to cover all of the municipalities that form the urban complex of Thessaloniki. Grids were selected based on accessibility and potential suitable habitat for RNP; however, four of the selected grid squares were ultimately not surveyed due to access restrictions or safety concerns.” In addition, the Results have been modified as well, line 224: “Of the 99 grid squares that were surveyed during both breeding periods, RNPs were detected at least once in 45 of them (Figure 2). Naïve occupancy (i.e. without accounting for imperfect detection) illustrated a 11.2% increase between survey years, from 34.3% in 2024, to 45.5% in 2025 it increased. At sites where the species was detected at least once, the mean observed detection frequency across surveys was 0.72 (range: 0.25 – 1). Similarly, observed abundance of RNP varied among sites and survey years. Across all surveyed grid squares, the maximum number of individuals recorded at a single site (i.e. based on the higher count from the two surveys conducted within each year) increased from 97 individuals in 2024 to 156 individuals in 2025 (Figure 2). A small number of sites, located within the city center, supported particularly large aggregations. At sites were RNP were detected at least once, the mean naïve abundance per site increased from 2.85 individuals (range: 1 – 15) in 2024 to 3.46 individuals (range: 1 – 21) in 2025.”

 

 

Comment 16

Table 2: What is this parameter for?

Response: This refers to the parameter estimates of the extinction probability (ε), were  = -10.5 (logit-scale) and which indicates an extinction probability of < 0.001. That is, extinction probability was estimated as effectively zero, as no local extinctions were observed between sampling years/primary periods. See also comment #14. Estimates of extinction probability have now been added in Line 251: “Over the same study period, the estimated probability of colonization (γ) was 8.63% ( = -2.36, S.E. = 1.14), while extinction probability (ε) was effectively zero ( = -10.5, S.E. = 33.6).”.

Comment 17

Line 263: Too many references.

Response: Agreed. References previously numbered [10], [16] [18], [19], [20], and [69] have been removed.

Comment 18

Line 294: Calculations based on two seasons are not representative.

Response: Thank you for highlighting this important issue. We agree with your core point that a growth rate calculated from only two years of data is not representative of a sustained, annual population trend. Ecological populations are dynamic, and a single interval may reflect temporary conditions. We will therefore reframe our results and discussion entirely, to avoid any implication that we have estimated a representative annual trend. Our model provides the best estimate of the finite population change for the specific, single transition from 2024 to 2025, while accounting for imperfect detection. The narrow confidence interval (1.37 – 1.97) indicates that this measured increase was strong and statistically significant for this period, but as you correctly pointed out, it is still just one data point in time. To address your concern directly, and prevent any misinterpretation, we have implemented the following revisions.

Line 18: In addition, their population showed strong short-term growth, increasing by an estimated 64% from 2024 to 2025.

Line 32: The proportion of occupied sites increased by more than 10% between survey years (2024 – 2025), while the estimated population growth rate for this interval was 1.64, signaling a substantial short-term increase.  

Line 198: RNP population changes between 2024 and 2025 were estimated using the models of Dail and Madsen; also known as dynamic N-mixture models, which generalize the N-mixture model by relaxing the population closure assumption.

Line 280: The estimated growth rate (λ) between survey years was 1.64 (1.37 – 1.97), reflecting a strong population increase of 64% from 2024 to 2025.

Line 294: The key findings of widespread occupancy and strong positive growth rate during our study period, highlight the remarkable ability of this non-native species to flourish in urban-dominated environments.

Line 334: Our two-year data indicate a substantial population increase, similar in magnitude to growth rates observed during establishment phases in other regions.

Line 377: Although there is currently no mandatory requirement to control RNP populations in Europe [11], their strong population increase observed during the study and high modeled growth potential may pose risks to native species mainly through competition for resources [18,22–25,27,91].   

Line 387: This pattern is also reflected in both its widespread distribution and the substantial population growth we documented within the urban complex of Thessaloniki between 2024 and 2025.

Comment 19

Line 316: Too many references.

Response: Agreed. References previously numbered [35] and [36] have been removed.

Comment 20

Line 336: Too many references.

Response: Agreed. References previously numbered [35] and [91] have been removed.

Reviewer 2 Report

Comments and Suggestions for Authors

The review of the work entitled “ Spatial and temporal trends in the invasion dynamics of the Ring-necked parakeet (Psittacula krameri) in the urban complex of Thessaloniki, Greece.  ” provided an opportunity to highlight an important issue concerning the development of invasive species populations. This is a very topical research topic that should be addressed in scientific studies. Therefore, I consider the manuscript to be of high importance. 
Overall, it is a well-written, properly cited work. It is supported by a broad review of sources and grounded in the current state of knowledge. I have some comments on the methodology—too little information was presented in the results. In particular, the basic results of the invader inventory are missing. It is a pity that the authors did not add a single paragraph summarizing the basic results, which would have increased the value of the work. Especially since references to unpresented data appear in the discussion. A decision must be made – either show the data and discuss it, or remove the topic from the discussion as well. 

My most serious comments concern the analytical part. In my opinion, it is not justified to assess such detailed parameters as extinction or colonization, etc., on the basis of four counts in two seasons. The authors should focus on a solid analysis of occurrence, or possibly the progress of abundance in this short period. The assessment of the abundance trend over a two-year period is of little value. The models and the interpretation of the terms and symbols used are unclear. This requires clarification and explanation in the methodology of how the models were constructed. The use of three types of distributions for the same data was very surprising. One should be chosen and the choice justified. 
Apart from methodological suggestions, I highly value the submitted work.

Detailed comments are provided below:

l. 33 - "Naïve occupancy increased by more" - what means "Naïve" ?

l. 34 - than 10% between survey years - this information is useless without information about years of field works

l.  35 - indicating rapid expansion - once again, years of study are needed

l. 36 - if "northern" is needed? Maybe in Greece?

l. 37-38 - it is to similiar with l. 21-22. Please delete this in simple summary

l. 41 - 7 keywords are prepared. I suggest to delete "rose-ringed parakeet" - this name is in title

l. 48 - "array" ?

l. 88 - "managers" - try to describe them more specific. Who is responsible for this task in Greece?

l. 93-98 - any hypothesis was formulated ?

l. 130 - "range: 492 – 2.075 m" - Quite a wide range. Couldn't it have affected the results?

l. 153-171 - I can't undestand the data presented in supplement tables. What means "(.)" or " "?

l. 153-155 - few more informations about creating the data about colonization or extinction have to be added. For now I don't understand how it was prepared. You collected presence-absence data from the field. How changed it at extinction or colonization between one year? It could be just natural fluctuations during colonization process.

l. 158-160 - So why is it modeled when it's zero? A parameter that serves no purpose and hinders the reception of the good parts of the work

l. 173 - two years that's a very short period of time to set trends. Try to nuanse this part

l. 184-185 - but what for different types of distribution were tested? Only appropriate  should be used. Please present the histogram of data distribution

l. 185-187 - the best (top ranked) model does not have to be the best. If the wrong distribution is used, the results are false positives.

l. 208-224 - please inform what "𝜷̂" means?

l. 231 - any evidence?

Table 2,3 - could You add the p-value for variables? 

l. 294-297 - I think, that this data should be presented in Results section. Some simple raw data from the field: How many pairs were found, how many at most per square. What were the results in 2024 and 2025, what was the rate of change for individual squares, etc. 

l. 298 - move to Result section

l. 304 - change "are" at "were"

Author Response

Response to Reviewer 2

The authors would like to thank Reviewer 2 for their time and effort invested in reviewing our manuscript. Their comments were very constructive and helped us improve our work.

Below, we provide a point-by-point response to the reviewer’s comments and suggestions.

All changes/additions are highlighted in the revised version of our manuscript.

General comment:

The review of the work entitled “Spatial and temporal trends in the invasion dynamics of the Ring-necked parakeet (Psittacula krameri) in the urban complex of Thessaloniki, Greece” provided an opportunity to highlight an important issue concerning the development of invasive species populations. This is a very topical research topic that should be addressed in scientific studies. Therefore, I consider the manuscript to be of high importance.

Overall, it is a well-written, properly cited work. It is supported by a broad review of sources and grounded in the current state of knowledge. I have some comments on the methodology—too little information was presented in the results. In particular, the basic results of the invader inventory are missing. It is a pity that the authors did not add a single paragraph summarizing the basic results, which would have increased the value of the work. Especially since references to unpresented data appear in the discussion. A decision must be made – either show the data and discuss it, or remove the topic from the discussion as well.

My most serious comments concern the analytical part. In my opinion, it is not justified to assess such detailed parameters as extinction or colonization, etc., on the basis of four counts in two seasons. The authors should focus on a solid analysis of occurrence, or possibly the progress of abundance in this short period. The assessment of the abundance trend over a two-year period is of little value. The models and the interpretation of the terms and symbols used are unclear. This requires clarification and explanation in the methodology of how the models were constructed. The use of three types of distributions for the same data was very surprising. One should be chosen and the choice justified.

Apart from methodological suggestions, I highly value the submitted work.

Response: We thank the reviewer for the careful reading of our manuscript and for recognizing its relevance and overall quality. We appreciate the positive comments regarding the importance of the topic, the clarity of presentation, and the study’s alignment with current scientific knowledge. All methodological remarks were considered in detail and the manuscript has been revised accordingly. Our detailed responses regarding each specific comment are provided below.

Comment 1

1. 33 - "Naïve occupancy increased by more" - what means "Naïve" ?

Response: Thank you for raising this important issue for clarity. “Naïve occupancy” is a standard term used in occupancy modeling and refers to the simple proportion of sites where the species was detected at least once during a survey period, without accounting for imperfect detection. Based on the above, naïve occupancy = the number of sites with at least one detection/the total number of surveyed sites. This contrasts with the estimated initial probability of occupancy (ψ₁) from the dynamic occupancy model, which is model-based and corrects for the probability of detection (p). We apologize for not clarifying this term initially. The issue has been clarified accordingly, to improve accessibility for all readers.

Line 32: “The proportion of occupied sites increased by more than 10% between survey years (2024 – 2025), while the estimated population growth rate for this interval was 1.64, signaling a substantial short-term increase.”

Line 225: “Naïve occupancy (i.e. without accounting for imperfect detection) illustrated a 11.2% increase between survey years, from 34.3% in 2024, to 45.5% in 2025 it increased.”

Comment 2

1. 34 - than 10% between survey years - this information is useless without information about years of field works

Response: Agreed. Survey years have been added accordingly in Line 32: The proportion of occupied sites increased by more than 10% between survey years (2024 – 2025), while the estimated population growth rate for this interval was 1.64, signaling a substantial short-term increase.

Comment 3

1. 35 - indicating rapid expansion - once again, years of study are needed

Response: Agreed. As with the previous comment, the sentence has been modified to include survey years. Line 32: The proportion of occupied sites increased by more than 10% between survey years (2024 – 2025), while the estimated population growth rate for this interval was 1.64, signaling a substantial short-term increase.

Comment 4

1. 36 - if "northern" is needed? Maybe in Greece?

Response: Thank you for making this suggestion. We have modified the sentence accordingly. Line 35: “Our findings provide the first detailed evidence of an established and growing RNP population within the urban complex of Thessaloniki, Greece.” We included the terms “urban complex of Thessaloniki”, since the inclusion of “Greece” alone could falsely imply national coverage.

Comment 5

22. 37-38 - it is to similiar with l. 21-22. Please delete this in simple summary

Response: Agreed. Lines have been deleted from the Simple Summary as suggested.

 

Comment 6

1. 41 - 7 keywords are prepared. I suggest to delete "rose-ringed parakeet" - this name is in title

Response: Agreed. The keyword has been deleted as it may lead to confusion regarding the studied species.

Comment 7

1. 48 - "array" ?

Response: To improve clarity and ensure accessibility for a broad readership, we have revised the wording by replacing the term array. The revised text now reads as follows Line 47: “While many introduced species fail to persist, some establish viable populations, whilst an even smaller portion is able to spread and cause diverse negative impacts [4].”

Comment 8

1. 88 - "managers" - try to describe them more specific. Who is responsible for this task in Greece?

Response: We thank the reviewer for this suggestion. In Greece, the responsibility for managing invasive species is often shared among multiple ministries, agencies and scientific bodies, reflecting the complexity of the issue. For specific cases the responsible body can vary, but overall, the Ministry of Environment and Energy is the primary coordinator (in accordance with EU Regulation 1143/2014) with operational roles distributed across specialized agencies, regional and local authorities. Based on the above, Line 87 has been modified as follows: “Given the species’ exponential increase in population numbers and rapid expansion potential [7], as well as the rising evidence of negative impacts on native fauna [31], managers; primarily the Ministry of Environment and Energy in Greece, together with regional authorities, Forestry Services, and protected-area management units, are often called upon to take effective measures aimed at reducing invasive species’ population numbers or controlling their spread [36].”

Comment 9

1. 93-98 - any hypothesis was formulated ?

Response: We thank the reviewer for this important observation. We have revised the Introduction to explicitly state our research hypotheses, now included in line 95: “In this context, our study offers the first detailed analysis of the spatial and temporal trends of RNP expansion in Greece, focusing on the urban complex of Thessaloniki. Based on previous studies regarding the species, we hypothesized that (i) RNP occupancy and abundance would be positively associated with urbanization intensity and greater availability of nesting and foraging sites, and (ii) the population would show an increasing trend between survey years, reflecting the species’ ongoing spread and expansion. Our study could provide insights into the mechanisms that facilitate RNP establishment and persistence in Mediterranean urban environments and could serve as a baseline for long-term monitoring of the species’ spread and population size, and for assessing the outcomes of management and control measures.”

Comment 10

2. 130 - "range: 492 – 2.075 m" - Quite a wide range. Couldn't it have affected the results?

Response: We thank the reviewer for raising this point regarding potential effects of transect length on model results. Variation in sampling effort was explicitly considered during model construction. Transect length was included as an observation-level covariate in both the dynamic occupancy and dynamic N-mixture models to account for potential effects on detection probability (see Table 1, Line 178: “Time of survey, date and year, as well as line transect length, were considered as potential covariates affecting detectability.”, as well as Line 207: “In the first stage, we built a set of candidate models to estimate p, using the same variables that could potentially affect detectability as in our dynamic occupancy models (Table 1) and explored three alternative statistical distributions; Poisson (P), Negative Binomial (NB), and Zero-Inflated Poisson (ZIP).” In both modeling frameworks, transect length showed no meaningful effect and was not retained in the top-ranked models based on model selection criteria. This indicates that, within the observed range of transect lengths, variation in transect length did not significantly influence detection estimates. Consequently, the covariate was excluded from the final models for parsimony (see Line 240: “Our top dynamic occupancy model (Table 2) showed that detection probability (p) was not influenced by any of the observation-level covariates.” and Line 269: “Detection probability (p) was not influenced by any of the observation-level covariates, with the null detection model providing the best fit.”)

Comment 11

1. 153-171 - I can't undestand the data presented in supplement tables. What means "(.)" or " "?

Response: We thank the reviewer for pointing out the need for clearer notation in our supplementary tables and we apologize for any confusion. The parentheses with a dot (.) are standard notation in ecological modeling. They indicate that a parameter was held constant (not modeled as a function of any covariates) for that particular model. To improve clarity, we have included a short explanation at the footing of all corresponding tables found in the supplementary materials, which reads: “*(.) indicates parameters that were held constant – no covariates effects”.

Comment 12

1. 153-155 - few more informations about creating the data about colonization or extinction have to be added. For now I don't understand how it was prepared. You collected presence-absence data from the field. How changed it at extinction or colonization between one year? It could be just natural fluctuations during colonization process.

Response: Colonization and extinction were not created or coded by us. They are latent processes estimated by the dynamic occupancy model from repeated presence/absence surveys across years, while accounting for imperfect detection. Presence/absence data were collected at each grid, with two visits per grid per year. We used dynamic occupancy models to analyze our data, which explicitly accounts for imperfect detection; such that changes in detection between years are not automatically interpreted as colonization or extinction, but are modelled probabilistically through the colonization (γ) and extinction (ε) parameters. Hence, apparent year to year changes may reflect either true changes in occupancy or variation in detection, and the model distinguishes between these processes.

Comment 13

1. 158-160 - So why is it modeled when it's zero? A parameter that serves no purpose and hinders the reception of the good parts of the work

Response: Thank you for this critical question. We agree that testing covariates for extinction (ε) is often a key part of dynamic occupancy analysis. However, in our study no sites that were occupied during the first year of surveys were detected as unoccupied in the second year. If we tried to fit a covariate to ε, model estimation would fail since optimization algorithms cannot estimate coefficients for a process not observed in the data. Therefore, we constrained ε(.) to be constant across all models. This is a standard and necessary approach in such cases, as it reflects the empirical reality of our data over the 2-year study period, and allows the model to focus on estimating the parameters for which we have information.  Moreover, it is not statistically possible to exclude the extinction parameter and still fit a dynamic occupancy model, since all parameters are jointly estimated. We have now clarified this critical point in line 181: “Finally, because no instances of local extinction were detected during the second survey year (i.e., all sites occupied in 2024 remained occupied in 2025), extinction probability (ε) was held constant throughout the modeling process. This constraint reflects the observed biological pattern during the study period and ensures model estimability.” In addition, in line 172 it is now clearly stated that all model parameters are jointly estimated: “RNP detection and non-detection data were analyzed using multi-season dynamic occupancy models, wherein initial occupancy (ψ₁; i.e. the probability that a site was occupied during the first year of surveys), colonization (γ; i.e. the probability that an unoccupied site becomes occupied between survey years), extinction (ε; i.e. the probability that an occupied site becomes unoccupied between survey years), and detectability (p; i.e the probability of detecting the species during a survey, conditional to its presence) are jointly estimated to describe state transitions across the surveyed area.”

Comment 14

1. 173 - two years that's a very short period of time to set trends. Try to nuanse this part

Response: Thank you for highlighting this important issue. We agree with your core point that a growth rate calculated from only two years of data is not representative of a sustained, annual population trend. Ecological populations are dynamic, and a single interval may reflect temporary conditions. We will therefore reframe our results and discussion entirely, to avoid any implication that we have estimated a representative annual trend. Our model provides the best estimate of the finite population change for the specific, single transition from 2024 to 2025, while accounting for imperfect detection. The narrow confidence interval (1.37 – 1.97) indicates that this measured increase was strong and statistically significant for this period, but as you correctly pointed out, it is still just one data point in time. To address your concern directly, and prevent any misinterpretation, we have implemented the following revisions.

Line 18: In addition, their population showed strong short-term growth, increasing by an estimated 64% from 2024 to 2025.

Line 32: The proportion of occupied sites increased by more than 10% between survey years (2024 – 2025), while the estimated population growth rate for this interval was 1.64, signaling a substantial short-term increase. 

Line 198: RNP population changes between 2024 and 2025 were estimated using the models of Dail and Madsen; also known as dynamic N-mixture models, which generalize the N-mixture model by relaxing the population closure assumption.

Line 280: The estimated growth rate (λ) between survey years was 1.64 (1.37 – 1.97), reflecting a strong population increase of 64% from 2024 to 2025.

Line 294: The key findings of widespread occupancy and strong positive growth rate during our study period, highlight the remarkable ability of this non-native species to flourish in urban-dominated environments.

Line 334: Our two-year data indicate a substantial population increase, similar in magnitude to growth rates observed during establishment phases in other regions.

Line 377: Although there is currently no mandatory requirement to control RNP populations in Europe [11], their strong population increase observed during the study and high modeled growth potential may pose risks to native species mainly through competition for resources [18,22–25,27,91].   

Line 387: This pattern is also reflected in both its widespread distribution and the substantial population growth we documented within the urban complex of Thessaloniki between 2024 and 2025.

Comment 15

1. 184-185 - but what for different types of distribution were tested? Only appropriate  should be used. Please present the histogram of data distribution
2. 185-187 - the best (top ranked) model does not have to be the best. If the wrong distribution is used, the results are false positives.

Response: We thank the reviewer for this critical and insightful comment. As mentioned in our manuscript, three different types of distributions were tested; Line 207: “In the first stage, we built a set of candidate models to estimate p, using the same variables that could potentially affect detectability as in our dynamic occupancy models (Table 1) and explored three alternative statistical distributions; Poisson (P), Negative Binomial (NB), and Zero-Inflated Poisson (ZIP).” Our model evaluation process involved two main steps: model selection/ranking using AIC and goodness-of-fit assessment of the top performing model. The Negative Binomial model was supported by the data based on AIC. In addition, and in order to address any concerns regarding the top model’s performance, we conducted a goodness-of-fit test using the parametric bootstrap approach. We simulated 1000 datasets from the fitted top ranked model and calculated Pearsons χ². The bootstrap p-value for the top-ranked model was 0.174, indicating no evidence of lack-of-fit. Therefore, our selection of the Negative Binomial model is supported by information-theoretic criteria (AIC), formal goodness-of-fit testing, and biological plausibility (i.e. the Negative Binomial distribution models overdispersion, which is ecologically realistic for our study species perhaps due to factors such as social aggregation, which is the tendency of individuals of a species to come together and form groups, clusters, or colonies based on social behaviour. This is a common ecological cause f clumped spatial distribution, which in turn is a primary biological reason why count data often follow a Negative Binomial distribution.

Comment 16

1. 208-224 - please inform what "?̂" means?

Response: Thank you for this question, which allows us to clarify the interpretation of our models’ parameters. In our models, ­ represents a regression coefficient, which is referred to as “parameter estimates”. It quantifies the direction and the strength of the relationship between a covariate (e.g. percent cover of green areas) and the log-transformed expected value of a biological parameter (e.g. initial abundance). In our results (Tables 2 and 3), we present parameter estimates ­ and highlight those variables with confidence intervals (85% and 95%) not spanning zero. The estimate itself gives the effect direction and size on the corresponding scale (logit/log), while confidence intervals indicate precision (see Comment #18). Based on the results from Tables 2 and 3, we have also plotted model predictions for the most informative variables (Figures 3 and 4), which show the relationships back-transformed to the biological scale. To address this issue, we have added the words “Parameter estimates (­)” in Tables 2 and 3.

Comment 17

1. 231 - any evidence?

Response: Please see Comment #15 as it refers to the same issue regarding alternative statistical distributions and how they were tested.

Comment 18

Table 2,3 - could You add the p-value for variables? 

Response: We thank the reviewer for their suggestion to include p-values for the model variables. However, dynamic occupancy and N-mixture models are complex hierarchical models that are most commonly evaluated using an information-theoretic approach. In such frameworks, the primary focus is on effect sizes and their precision (i.e. parameter estimates and confidence intervals), rather than binary hypothesis testing via p-values. As stated in our manuscript, the strength of evidence for a variable is represented by whether its 85% or 95% confidence intervals span zero (see Line 218: “For both the dynamic occupancy and the dynamic N-mixture models, parameter estimates with 95% confidence intervals not spanning zero were interpreted as highly informative, those with 85% confidence intervals not spanning zero were interpreted as moderately informative, and those with 85% confidence intervals spanning zero were considered uninformative [65].”). A p-value from a single global model would not reflect this this multi-model inference process. On the other hand, presenting estimates and confidence intervals allows readers to assess both the statistical significance (via confidence intervals spanning or not zero) and the biological significance (magnitude of effect), which is often more important for conservation and management application. Based on the above rational, we have decided not to include p-values as suggested. Instead, Tables 2 and 3 include a footnote explicitly stating which parameters/variables have confidence intervals not spanning zero at the 85% and 95% level.

Comment 19

1. 294-297 - I think, that this data should be presented in Results section. Some simple raw data from the field: How many pairs were found, how many at most per square. What were the results in 2024 and 2025, what was the rate of change for individual squares, etc. 

Response: Agreed. The line has been modified accordingly (Line 334: “Our two-year data indicate a substantial population increase, similar in magnitude to growth rates observed during establishment phases in other regions.”). In addition, and based on the reviewer’s insightful suggestion, we have added some additional descriptive results in Line 224: “Of the 99 grid squares that were surveyed during both breeding periods, RNPs were detected at least once in 45 of them (Figure 2). Naïve occupancy (i.e. without accounting for imperfect detection) illustrated a 11.2% increase between survey years, from 34.3% in 2024, to 45.5% in 2025 it increased. At sites where the species was detected at least once, the mean observed detection frequency across surveys was 0.72 (range: 0.25 – 1). Similarly, observed abundance of RNP varied among sites and survey years. Across all surveyed grid squares, the maximum number of individuals recorded at a single site (i.e. based on the higher count from the two surveys conducted within each year) increased from 97 individuals in 2024 to 156 individuals in 2025 (Figure 2). A small number of sites, located within the city center, supported particularly large aggregations. At sites were RNP were detected at least once, the mean naïve abundance per site increased from 2.85 individuals (range: 1 – 15) in 2024 to 3.46 individuals (range: 1 – 21) in 2025.” Finally, we made some changes to our previous results, to make them clearer and easily understood (Line 243: “The estimated probability of initial occupancy (ψ) across the study area was 25.5% ( = -1.07, S.E. = 0.338).”, Line 251: “Over the same study period, the estimated probability of colonization (γ) was 8.63% ( = -2.36, S.E. = 1.14), while extinction probability (ε) was effectively zero ( = -10.5, S.E. = 33.6). Colonization probability tended to decrease with increasing distance from arable land (Dis_Arable:  = -2.24, S.E. = 1.84), although this effect was uninformative.”, Line 271: “The model estimated a mean initial abundance (λ₁) of 0.62 individuals per grid square in the first survey year ( = -0.472, S.E. = 0.255), representing baseline abundance at average covariate values.”, and Line 280: “The estimated growth rate (λ) between survey years was 1.64 (1.37 – 1.97), reflecting a strong population increase of 64% from 2024 to 2025.”)

 

 

 

Comment 20

1. 298 - move to Result section

Response: Agreed. Has been modified accordingly (Line 338: “Our results indicate a substantial RNP population within the study area, although the actual number of individuals is likely higher.”).

Comment 21

1. 304 - change "are" at "were"

Response: Thank you for your comment. The sentence has been corrected accordingly. Line 343: “Furthermore, the relatively low detection probability (~40%) estimated by the N-mixture model may also imply that not all individuals were detected during counts.”

Reviewer 3 Report

Comments and Suggestions for Authors General comments   The manuscript investigates the spatial and temporal dynamics of the Ring-necked parakeet (Psittacula krameri) based on a two-year survey, applying dynamic occupancy models and Dynamic N-mixture models to explore its association with urban environmental variables. The topic is relevant and timely, and the findings may provide useful insights for the management of invasive species in urban environments. However, several substantial issues limit the robustness of the conclusions. First, the sampling design and survey implementation lack sufficient clarity, which raises concerns about data representativeness and reliability. Second, the relatively short temporal span and the spatial selection of survey units constrain the strength of inferences about invasion dynamics. Finally, the manuscript contains a number of formatting and writing errors that require careful revision. Overall, while the study addresses a meaningful ecological question and employs appropriate analytical frameworks, improvements in methodological transparency, rigor, and interpretation are necessary to enhance the quality and credibility of the manuscript.
 

Specific comments

Line 41: The common name of the species is inconsistent in the keywords (“rose-ringed parakeet”) compared with the title and main text (“Ring-necked parakeet”). Please standardize the species name throughout the manuscript.

Line 66: Check for content errors, the '85.120 individuals' is incorrect.   Line 126: The selected unit grid for the study is based on accessibility and potential habitats. This is based on the preference settings of RNP for densely populated urban areas mentioned earlier. Will this cause some bias in the analysis results? Have the species survival status of RNP in remote suburban areas been considered? The reference in the following discussion about the unfavorable survival of RNP at low temperatures, considering the mild climate in the study area mentioned in the article, cannot dispel the possibility of sample representativeness issues in this part.   Line 121: Regarding the line transect surveys, there is a significant difference between the shortest and longest transect. How can the research ensure the possible deviation caused by the different lengths of each transect? The time period used in the study is the time at dawn and dusk, when the bird calls are complex. Researchers may also have error in sampling results within a time window that lasts for more than two hours. How does the research ensure accuracy?   Line 128: We agree to use the number of visually detected individuals in the line transect surveys for statistical analysis. But how is the number of RNP individuals heard defined? For example, how are multiple calls from one RNP recognized? As far as I know, the time period used in the study is the active period of bird chirping activity, usually with a lot of chirping sounds, which is not conducive to identifying the number of individuals. This raises doubts about the accuracy of the research method. The description of the line transect surveys in the previous section needs a more detailed introduction.   Line 130: Please verify numerical and unit-related errors (e.g., “1,184 m” and “2,075 m”) and check for similar issues throughout the manuscript.   Line 135: There is a clear formatting error (“9h300”). Please ensure the correct format of content.   Line 145: Given that each sampling unit represents a 1 km² grid, please provide the original spatial resolution of all GIS-derived environmental variables in a table or supplementary material. This information is essential for evaluating potential uncertainties introduced during resampling and aggregation.   Line 152: Dynamic occupancy models are typically designed for datasets with multiple consecutive temporal periods to reliably estimate colonization and extinction dynamics. In this study, the model is based on two years of data, with surveys conducted twice per site within a single month. This limited temporal replication raises concerns about whether the available data are sufficient to support robust inference on invasion dynamics and temporal trends. Furthermore, the number of repeat surveys per site is not clearly described, making it difficult to assess model adequacy.We have noticed that previous studies have certain requirements for sample size (McKann et al., 2013; Abrha et al., 2023). Therefore, the research should strengthen the rationality of using the model under these constraints.   Mckann, P. C., Gray, B. R., & Thogmartin, W. E. (2013). Small sample bias in dynamic occupancy models. The Journal of Wildlife Management, 77(1), 172–180. https://doi.org/10.1002/jwmg.433

Abrha, A. M., Gedeon, K., Podsiadlowski, L., & Töpfer, T. (2023). Dynamic occupancy modeling of a cryptic ground-dwelling pheasant species in the Upper Blue Nile Basin in Ethiopia. Global Ecology and Conservation, 48, e02710. https://doi.org/10.1016/j.gecco.2023.e02710

Line 159: We believe that the description of the extinction probability remaining constant during the two periods in the study requires more support. "no grid square was observed to be abandoned" does not directly imply that the extinction probability remains constant.

Line 294: The use of the term “annual” is not recommended, as the temporal coverage of the study does not fully support annual-scale inference. Please revise accordingly.

Author Response

Response to Reviewer 3

The authors would like to thank Reviewer 3 for their time and effort invested in reviewing our manuscript. Their comments were very constructive and helped us improve our work.

Below, we provide a point-by-point response to the reviewer’s comments and suggestions.

All changes/additions are highlighted in the revised version of our manuscript.

General comment:

The manuscript investigates the spatial and temporal dynamics of the Ring-necked parakeet (Psittacula krameri) based on a two-year survey, applying dynamic occupancy models and Dynamic N-mixture models to explore its association with urban environmental variables. The topic is relevant and timely, and the findings may provide useful insights for the management of invasive species in urban environments. However, several substantial issues limit the robustness of the conclusions. First, the sampling design and survey implementation lack sufficient clarity, which raises concerns about data representativeness and reliability. Second, the relatively short temporal span and the spatial selection of survey units constrain the strength of inferences about invasion dynamics. Finally, the manuscript contains a number of formatting and writing errors that require careful revision. Overall, while the study addresses a meaningful ecological question and employs appropriate analytical frameworks, improvements in methodological transparency, rigor, and interpretation are necessary to enhance the quality and credibility of the manuscript.

Response: We thank the reviewer for the overall evaluation and for highlighting the relevance of the topic. We have revised our manuscript according to the reviewer’s remarks. We acknowledge the concerns imposed by the two-year survey period and the urban focused sampling design, and have adjusted the interpretation of our results accordingly. In addition, the manuscript has been carefully edited to address formatting and language issues. Below we provide a point-by-point response to each specific comment.

Comment 1

Line 41: The common name of the species is inconsistent in the keywords (“rose-ringed parakeet”) compared with the title and main text (“Ring-necked parakeet”). Please standardize the species name throughout the manuscript.

Response: We apologize for the confusion. The keyword has been removed.

Comment 2: Line 66: Check for content errors, the '85.120 individuals' is incorrect.

Response: Agreed. Has been corrected.

Comment 3

Line 126: The selected unit grid for the study is based on accessibility and potential habitats. This is based on the preference settings of RNP for densely populated urban areas mentioned earlier. Will this cause some bias in the analysis results? Have the species survival status of RNP in remote suburban areas been considered? The reference in the following discussion about the unfavorable survival of RNP at low temperatures, considering the mild climate in the study area mentioned in the article, cannot dispel the possibility of sample representativeness issues in this part.

Response: We thank the reviewer for raising this important point regarding potential sampling bias. Grid selection was guided by accessibility and the presence of potential suitable habitat, reflecting the known ecological preference of RNP for densely populated urban environments. As such, our sampling design was not intended to represent the full suburban-rural gradient, but rather to characterize occurrence and abundance patterns within the urban complex of Thessaloniki, where the problem is more acute and management concerns are most relevant. Suburban areas where therefore not targeted and our results should be interpreted as primarily applying to urban and peri-urban settings, rather to marginal habitats. However, we do acknowledge the fact that RNP may occur (perhaps at lower densities) in less urbanized areas (Line 340: “In fact, several individuals were observed up to ~10 km beyond the survey boundaries (near the thermal springs of Lagkadas, personal observation), indicating that the species may be more abundant and widespread than our data suggest.”), and recommend that future studies should extend beyond the urban core (Line 380: “To anticipate and mitigate these risks, regular monitoring of population status is essential (Pârâu et al., 2016), alongside targeted research on ecological impacts, and should be extended to suburban and rural areas beyond the urban core.”).

Comment 4

Line 121: Regarding the line transect surveys, there is a significant difference between the shortest and longest transect. How can the research ensure the possible deviation caused by the different lengths of each transect? The time period used in the study is the time at dawn and dusk, when the bird calls are complex. Researchers may also have error in sampling results within a time window that lasts for more than two hours. How does the research ensure accuracy?

Response: We thank the reviewer for raising this point regarding potential effects of transect length and time of survey on model results. Variation in sampling effort and sampling timing was explicitly considered during model construction. Transect length and time of survey (expressed as minutes after sunrise) were included as  observation-level covariates in both the dynamic occupancy and dynamic N-mixture models to account for potential effects on detection probability (see Table 1, Line 178: “Time of survey, date and year, as well as line transect length, were considered as potential covariates affecting detectability.”, as well as Line 207: “In the first stage, we built a set of candidate models to estimate p, using the same variables that could potentially affect detectability as in our dynamic occupancy models (Table 1) and explored three alternative statistical distributions; Poisson (P), Negative Binomial (NB), and Zero-Inflated Poisson (ZIP).” In both modeling frameworks, transect length and time of survey showed no meaningful effect and were not retained in the top-ranked models based on model selection criteria. This indicates that, within the observed range of transect lengths, variation in transect length did not significantly influence detection estimates. Similarly, time of survey did not significantly influence detection estimates. Consequently, the covariates were excluded from the final models for parsimony (see Line 240: “Our top dynamic occupancy model (Table 2) showed that detection probability (p) was not influenced by any of the observation-level covariates.” and Line 269: “Detection probability (p) was not influenced by any of the observation-level covariates, with the null detection model providing the best fit.”)

Comment 5

Line 128: We agree to use the number of visually detected individuals in the line transect surveys for statistical analysis. But how is the number of RNP individuals heard defined? For example, how are multiple calls from one RNP recognized? As far as I know, the time period used in the study is the active period of bird chirping activity, usually with a lot of chirping sounds, which is not conducive to identifying the number of individuals. This raises doubts about the accuracy of the research method. The description of the line transect surveys in the previous section needs a more detailed introduction.

Response: We thank the reviewer for raising this important issue. We fully acknowledge that counting individual birds based on their calls; especially when clustered, is very challenging and may lead to over- or under-estimation if calls from the same individual are counted multiple times or vise-versa. RNP individuals were counted only when we could spatially and temporally isolate a calling event to a specific location and/or visually verify its presence and numbers. If multiple calls came unambiguously from the same location (and visual verification was not possible), they were attributed to a single individual. In addition, by moving along each transect we could spatially and temporally differentiate between potential different sources. Crucially, we recognize that our raw counts (or even our presence/absence data) are not claimed to show the true abundance (or true occupancy). This is precisely why we employed N-mixture (as well as dynamic occupancy) models. These models explicitly account for the imperfect detection process. Even so, and based on the results of our analysis, we acknowledge this issue as a possible limitation in abundance estimations and recommend conducting roost counts as an alternative (Line 349: “Hence, conducting roost counts, particularly in winter when deciduous trees have reduced foliage, may improve detectability and produce more accurate abundance estimates, given that all major roost sites are previously identified.”). Based on the reviewer’s comment, we have added additional information regarding RNP surveys, in Line 138: “RNP auditory cues primarily guided visual searches. When visual confirmation was not possible, vocalizations originating from a single point source were attributed to one individual”.

Comment 6

Line 130: Please verify numerical and unit-related errors (e.g., “1,184 m” and “2,075 m”) and check for similar issues throughout the manuscript.

Response: Agreed. Have been corrected accordingly.

 

 

 

Comment 7

Line 135: There is a clear formatting error (“9h300”). Please ensure the correct format of content.

Response: Agreed. Has been modified accordingly.

Comment 8

Line 145: Given that each sampling unit represents a 1 km² grid, please provide the original spatial resolution of all GIS-derived environmental variables in a table or supplementary material. This information is essential for evaluating potential uncertainties introduced during resampling and aggregation.

Response: Thank you for highlighting this important issue. Indeed, up-scaling of environmental variables could in some cases lead to potential uncertainties. However, in our study variables were available at a 1 km² resolution and even finer. While for most variables, their resolution is mentioned within the corresponding reference, we agree with the reviewer that this must be more easily verified. As such, we have included a new column in Table S1 (supplementary material), which shows the type and spatial resolution of the original dataset.

Comment 9

Line 152: Dynamic occupancy models are typically designed for datasets with multiple consecutive temporal periods to reliably estimate colonization and extinction dynamics. In this study, the model is based on two years of data, with surveys conducted twice per site within a single month. This limited temporal replication raises concerns about whether the available data are sufficient to support robust inference on invasion dynamics and temporal trends. Furthermore, the number of repeat surveys per site is not clearly described, making it difficult to assess model adequacy. We have noticed that previous studies have certain requirements for sample size (McKann et al., 2013; Abrha et al., 2023). Therefore, the research should strengthen the rationality of using the model under these constraints.   Mckann, P. C., Gray, B. R., & Thogmartin, W. E. (2013). Small sample bias in dynamic occupancy models. The Journal of Wildlife Management, 77(1), 172–180. https://doi.org/10.1002/jwmg.433

Abrha, A. M., Gedeon, K., Podsiadlowski, L., & Töpfer, T. (2023). Dynamic occupancy modeling of a cryptic ground-dwelling pheasant species in the Upper Blue Nile Basin in Ethiopia. Global Ecology and Conservation, 48, e02710. https://doi.org/10.1016/j.gecco.2023.e02710

Response: We thank the reviewer for this important comment regarding temporal replication and survey effort. We agree that dynamic occupancy models are ideally supported by larger time series and multiple repeat surveys per season. A greater number of repeated surveys per year would generally improve the precision of parameter estimates; particularly for detection probability, and would perhaps reduce issues of parameter identifiability (narrower confidence intervals). We acknowledge that for a cryptic or rare species, two visits per year could be insufficient to reliably separate occupancy from detection. However, the specific biological context of our study system; a rapidly expanding invasive species with conspicuous behaviour, makes a two-visit study both appropriate and informative. Simulation studies have shown that two-repeat surveys per site can provide adequate estimations when detection probability is moderate to high (Field et al., 2005; MacKenzie et al., 2002; Tyre et al., 2003). The RNP is a conspicuous and vocal species in urban environments and detection probability in our study was high (~80%), supporting the use of two-repeat surveys combined with a larger number of surveyed sites. Our study design represents a practical compromise, given the logistical and financial constraints inherent in multi-sites surveys. Our study prioritized spatial over temporal replication. This is a strategic choice when the primary goal is to understand landscape-scale patterns. It has been recommended that when detection probability is high, surveying more sites is generally better than increasing the number of surveys per site (Tyre et al., 2003). Our study represents a short-term assessment and is intended to estimate state transitions over a single interval rather than long-term invasion dynamics. Extinction probability was therefore held constant (see Comment #10) and population change was interpreted as a finite change between 2024 and 2025 (see Comment #11). Based on the above, and despite the limited number of repeated surveys, our study provides the first empirical estimates of RNP occupancy and abundance across an urban complex and can serve as a crucial baseline for future monitoring informing on management prioritization. To address the reviewer’s concerns, we have made the following revisions:

Line 142: “Surveys were conducted during two consecutive primary sampling periods, corresponding to May 2024 and May 2025. Within each primary period, each transect was surveyed twice, resulting in two repeat surveys per site per year. Repeated surveys were carried out within the same month to satisfy the assumption of population closure within primary periods. Surveys were scheduled in two daily time windows (06:30 – 09:30 and 17:00 – 19:30) and each transect was surveyed once in each time window per year, ensuring temporal variation in detectability was accounted for.”

Line 150: “This level of within-season replication represents a deliberate trade-off between temporal and spatial sampling effort. Given the conspicuous and vocal behaviour of RNP in urban environments and the moderate to high detection probability observed in our study, two-repeat surveys per site are considered sufficient and can provide reliable estimates (Field et al., 2005; MacKenzie et al., 2002), while allowing broader spatial coverage (Tyre et al., 2003).”

Comment 10

Line 159: We believe that the description of the extinction probability remaining constant during the two periods in the study requires more support. "no grid square was observed to be abandoned" does not directly imply that the extinction probability remains constant.

Response: Thank you for this critical question. We agree that testing covariates for extinction (ε) is often a key part of dynamic occupancy analysis. However, in our study no sites that were occupied during the first year of surveys were detected as unoccupied in the second year. If we tried to fit a covariate to ε, model estimation would fail since optimization algorithms cannot estimate coefficients for a process not observed in the data. Therefore, we constrained ε(.) to be constant across all models. This is a standard and necessary approach in such cases, as it reflects the empirical reality of our data over the 2-year study period, and allows the model to focus on estimating the parameters for which we have information.  Moreover, it is not statistically possible to exclude the extinction parameter and still fit a dynamic occupancy model, since all parameters are jointly estimated. We have now clarified this critical point in line 181: “Finally, because no instances of local extinction were detected during the second survey year (i.e., all sites occupied in 2024 remained occupied in 2025), extinction probability (ε) was held constant throughout the modeling process. This constraint reflects the observed biological pattern during the study period and ensures model estimability.” In addition, in line 172 it is now clearly stated that all model parameters are jointly estimated: “RNP detection and non-detection data were analyzed using multi-season dynamic occupancy models, wherein initial occupancy (ψ₁; i.e. the probability that a site was occupied during the first year of surveys), colonization (γ; i.e. the probability that an unoccupied site becomes occupied between survey years), extinction (ε; i.e. the probability that an occupied site becomes unoccupied between survey years), and detectability (p; i.e. the probability of detecting the species during a survey, conditional to its presence) are jointly estimated to describe state transitions across the surveyed area.”

Comment 11

Line 294: The use of the term “annual” is not recommended, as the temporal coverage of the study does not fully support annual-scale inference. Please revise accordingly.

Response: Thank you for highlighting this important issue. We agree with your core point that a growth rate calculated from only two years of data is not representative of a sustained, annual population trend. Ecological populations are dynamic, and a single interval may reflect temporary conditions. We will therefore reframe our results and discussion entirely, to avoid any implication that we have estimated a representative annual trend. Our model provides the best estimate of the finite population change for the specific, single transition from 2024 to 2025, while accounting for imperfect detection. The narrow confidence interval (1.37 – 1.97) indicates that this measured increase was strong and statistically significant for this period, but as you correctly pointed out, it is still just one data point in time. To address your concern directly, and prevent any misinterpretation, we have implemented the following revisions.

Line 18: In addition, their population showed strong short-term growth, increasing by an estimated 64% from 2024 to 2025.

Line 32: The proportion of occupied sites increased by more than 10% between survey years (2024 – 2025), while the estimated population growth rate for this interval was 1.64, signaling a substantial short-term increase. 

Line 198: RNP population changes between 2024 and 2025 were estimated using the models of Dail and Madsen; also known as dynamic N-mixture models, which generalize the N-mixture model by relaxing the population closure assumption.

Line 280: The estimated growth rate (λ) between survey years was 1.64 (1.37 – 1.97), reflecting a strong population increase of 64% from 2024 to 2025.

Line 294: The key findings of widespread occupancy and strong positive growth rate during our study period, highlight the remarkable ability of this non-native species to flourish in urban-dominated environments.

Line 334: Our two-year data indicate a substantial population increase, similar in magnitude to growth rates observed during establishment phases in other regions.

Line 377: Although there is currently no mandatory requirement to control RNP populations in Europe [11], their strong population increase observed during the study and high modeled growth potential may pose risks to native species mainly through competition for resources [18,22–25,27,91].   

Line 387: This pattern is also reflected in both its widespread distribution and the substantial population growth we documented within the urban complex of Thessaloniki between 2024 and 2025.

 

References:

Field, S. A., Tyre, A. J., & Possingham, H. P. (2005). Optimizing allocation of monitoring effort under economic and observational constraints. The Journal of Wildlife Management, 69(2), 473–482.

MacKenzie, D. I., Nichols, J. D., Lachman, G. B., Droege, S., Andrew Royle, J., & Langtimm, C. A. (2002). Estimating site occupancy rates when detection probabilities are less than one. Ecology, 83(8), 2248–2255.

Tyre, A. J., Tenhumberg, B., Field, S. A., Niejalke, D., Parris, K., & Possingham, H. P. (2003). Improving precision and reducing bias in biological surveys: estimating false‐negative error rates. Ecological Applications, 13(6), 1790–1801.

Round 2

Reviewer 3 Report

Comments and Suggestions for Authors

The authors have provided a thorough and careful revision, and the major concerns raised during the review process have been adequately addressed. The review mainly raised three types of issues: some minor errors, doubts about research methods, and the need to supplement sample information. The author responded to these three main aspects.

1. Formatting and writing issues have been corrected.
2. More explanations were provided regarding the questioning of research methods. Explained the recognition of bird vocalizations and issues with the model, and added reference explanations where necessary.
3. Relevant explanations have been added to the manuscript regarding sample size, study area, and temporal range.

Overall, the authors’ efforts have improved the logical structure and methodological rigor of the manuscript. While the current study is suitable for publication, future research could further strengthen the conclusions by expanding the spatial and temporal scope and increasing sample size to explore invasion dynamics in greater detail.