The Following Report was posted on March 1, 2017 - Photos and graphs would not transfer from the original PDF. If you would like a complete copy of this report please email firstname.lastname@example.org and we will email you a copy.
The Red-crowned Parrot in South Texas:
Roost dynamics and annual reproductive output
Interim report on selected activities June 2016 - January 2017
TPWD Contract Number – 484290
AgriLife Contract number – M1602920
CFDA Number – 15.634
Department of Wildlife and Fisheries Sciences
Texas A&M University
Donald J. Brightsmith
Schubot Exotic Bird Health Center
Department of Veterinary Pathobiology
Texas A&M University
Determining annual reproductive output of Red-crowned Parrots (Amazona viridigenalis):
Knowledge of reproduction is important for assessing species status and interpreting population fluctuations. Monitoring reproductive effort of Red-crowned Parrots in an urban landscape is a daunting task. Finding and monitoring individual nests for a species that can travel tens of miles per day is difficult, costly, and time consuming. However, by documenting the proportion of juveniles in the population during roost counts we can estimate annual reproductive output and better understand seasonal population changes.
Although documenting juveniles at roosts may be a simpler method of analyzing reproductive output, it doesn’t mean it is easy. The spatial and temporal dynamics of roosts and an urban landscape make obtaining good looks and photographs of these parrots a difficult task. The behavior of the parrots and the lack of good lighting during evening roosts doesn’t help either. Our time span of getting juvenile data is limited; juveniles start showing up at roosts around mid-June and peak during late July and August.
We make sure to be in roosting areas a minimum of 45 minutes before sunset (usually closer to an hour) to give a chance of finding birds during their pre-roost. Parrots are usually foraging and socializing before they finally settle for the evening which gives us a good opportunity to find them with good lighting. This allows us to obtain photographs (sometimes) and to analyze behavior and size of family groups before the parrots slink into the dense foliage of Live Oak and Eucalyptus trees for the night.
In the valley, Brownsville is the easiest of the four roosts to find the parrots. It is the only roost where the birds have been seen on 100% of our 12 visits. The other three roosts - Harlingen, Weslaco, and McAllen - all have a larger area (multiple neighborhoods) where the birds will convene before roosting, which means we spend a lot of time driving around looking for them while losing valuable sunlight.
Once the parrots are found, we stick with them until they settle in for the evenings roost. While watching them, we take note of the group sizes they are flying in as well as whether they are an adult or juvenile. Juveniles can be told from adults by a combination of attributes including begging behavior, limited red on forehead, short tail length, and dark iris (Fig. 1). Each time we observed a roost we checked on average 26% ± 21% (N = 18 nights) of the birds to see if they were young or adults and we are working to get better at quickly and accurately identifying them.
Figure 1: Juvenile and adult Red-crowned Parrots (Amazona viridigenalis). “A” shows a juvenile RCPA – note dark iris and limited red on forehead. “B” is an adult RCPA – note golden iris and red extending behind center of eye on crown. “C” is a juvenile RCPA (right) being fed by an adult – note the behavior and short, jagged tail of the juvenile. “D” shows an adult (front) and juvenile – note eye color, amount of red on forehead, and overall impression.
The earliest fledglings we detected at the roosts in 2016 were in Weslaco on June 19th, McAllen June 20th, Harlingen June 23rd, and Brownsville July 11th. From June through August we counted adults and juveniles at roosts on 18 occasions. During these 18 nights, we made 539 checks of individual birds. Of these, 135 showed signs that they were young of the year (had fledged within the last 2 months). A total of 404 birds checked appeared to be post first year birds (> 12 months). These results suggest that about 25% of the birds coming to roost during July and August 2016 were juveniles. Of course, this estimate is not without bias. We may have biased our counts upward somewhat because begging juveniles often caught the attention of the observers. In addition, the 539 birds checked undoubtedly include repeats, meaning that a subset of the same individuals was likely checked on different nights. Within a night, we are quite confident that we did not recount the same individuals. Additional analyses are needed to better determine the average number of young of the year at the roost and the uncertainties around these estimates.
We thought that the percentage of young might vary among the different cities, suggesting that some cities were more suitable for the parrots than others. However, the estimated percent young was surprisingly consistent among all four roosts (Table 1). Our counts ended on August 12th before the numbers of young at the roosts began to decline. This prevented us from having a full fledging seasons worth of data and we plan to fill in that gap this year.
# nights juveniles checked
Number of birds checked
Raw % of juveniles
44% but unreliable due to small sample size
The finding that 25% of the young RCPAs coming to roost are young of the year suggests extremely high reproductive success in this population. In two other locations, research teams have used this same methodology to estimate the percent young in other populations of large Neotropical parrots. Matuzak and Brightsmith (2007) estimated that about 12.5% of the Yellow-naped Parrots (Amazona auropalliata) coming to roost in a population in Costa Rica were young of the year. Brightsmith (unpublished data) has estimated that 11.4% ± 1.3% (N= 5 years) of the large macaws (Ara macao, A. ararauna, and A. chloropterus) visiting a clay lick in southeastern Peru were young of the year.
Table 1: Percentage of juvenile versus adult Red-crowned Parrots at all four roost locations in the Lower Rio Grande Valley from June 19th to August 11th 2016.
In Costa Rica and Peru we had no evidence that the populations were increasing. However, Christmas Bird count data suggest that the population of RCPAs in the LRGV has been growing rapidly since about 1993. Our data on the large number of young RCPAs coming to roost strongly support the conclusion that the recent increase in population of RCPs in the LRGV has been due to high levels of reproduction and is not due to a continued release of captive birds or immigration from Mexico. This consistently high level of reproduction occurring across the range of the species in the LRGV suggests that, at least for the short term, the LRGV population of this species should continue to increase at a high rate. How long this rapid growth will continue is uncertain as at some point, the population will approach carrying capacity and either food or nesting sites will become limiting.
Roost attendance and seasonal dynamics of Red-crowned Parrots:
Data suggests that Red-crowned Parrot attendance at roost sites can vary greatly from season to season and even from night to night. What causes these fluctuations and what do they mean?
Seasonal abundance patterns at roosts can be explained by the breeding phenology of Red-crowned Parrots. In the Texas winter from November to early February, Red-crowned Parrots flock together in large congregations. It is likely this is a function of searching for ideal food sources, building
Figure 2: Numbers of Red-crowned Parrots reported from the Lower Rio Grande Valley of Texas on Christmas Bird Counts 1976 – 2015.
and strengthening social relationships, and protection from predators. At roosts, the birds come in large groups numbering in the hundreds. We have evidence that roosts are very dynamic at this time of year. During December 2016 and January 2017, the numbers of birds at each of the four major roosts fluctuated greatly. The numbers of birds roosting in Weslaco dropped from around 200 to near zero just as the roost in McAllen experienced a major increase (Fig. 3). Similarly, while the roost at Harlingen had no birds, the Brownsville roost shot up to about 300 birds (Fig. 4). These data suggest that the Harlingen roost combined with the Brownsville roost at least periodically over a span of about four weeks, and that the Weslaco roost combined with McAllen’s roost for about three weeks. Interestingly for the Weslaco and Brownsville roosts the numbers were quite similar before and after the fusion and fission events. It is unknown if individual birds shift regularly among roosts or if the large roosting groups move as units, but these data strongly suggest that there is movement among roosts. As a result, we suggest that all should work under the assumption that all the LRGV Red-crowned Parrots belong to a single large population, unless additional evidence causes us to doubt this assertion.
Figure 3: Maximum counts of Red-crowned Parrots at evening roosts in Weslaco and McAllen, Texas from January 15th 2016 to January 18th 2017. Figure includes Texas Parks and Wildlife valley wide surveys and counts by Simon Kiacz of Texas A&M.
Many food sources (Pecans, Live Oak acorns, Chinaberry etc) are available for parrots in the LRGV. Most of these food sources are available for a limited amount of time each season, so it may benefit the parrots to flock together to take advantage of seasonally abundant food sources. We have seen large flocks of parrots feeding in Pecan groves, large Chinaberry trees, and in neighborhoods full of fruiting Live Oaks.
Socializing is obviously an important part of parrot life, and perhaps this is why we as humans find them so interesting and entertaining. During the winter months, it is likely the parrots are using this time to develop relationships for the upcoming breeding season. Juveniles are likely refining their flying skills and learning how to forage on their own during this time, as well as forming relationships with other groups of parrots.
Safety in numbers is likely an important reason these birds roost together most nights. The ability of hundreds of pairs of eyes to find predators is a large incentive to fly a few extra miles every day. It makes sense to carry this behavior throughout the day while not protecting nest cavities or raising young.
The end of winter is the start of the breeding season for Red-crowned Parrots and we see a shift in their behavior. The large groups are seen less often as pairs split off into smaller groups, and as March comes around the pairs start to defend nest cavities in lieu of flocking and searching for food. We notice this in their behavior at roosts; the birds come in over a longer period of time and in smaller groups of two to 15 instead of hundreds at a time.
Figure 4: Maximum counts of Red-crowned Parrots at evening roosts in Harlingen and Brownsville from January 15th 2016 to January 18th 2017. Figure includes Texas Parks and Wildlife valley wide surveys. Figure includes Texas Parks and Wildlife valley wide surveys and counts by Simon Kiacz of Texas A&M.
Red-crowned Parrots start laying eggs at the beginning of April, and we notice a corresponding change at the roosts. Females stay at their nests to incubate eggs throughout the night and as a result we see a large drop in roost attendance (Figs. 3 and 4). Over the next few months, as nests start to fail and eggs start to hatch, we start to see an increase in numbers of birds at the evening roosts.
During mid-June, after roughly 27 days of incubating and 53 days as nestlings, the young start to fledge. Roosts become a cacophony of begging and feeding, as fledglings chase parents always begging for food. This is when we see the final uptick in roost attendance for the year, and it continues until fledging is finished during the beginning of August. From this point on we see only nominal changes in roost attendance until the large flocks start to appear again in winter.
Short term variability in roost counts
From June to August 2016 we conducted counts on sequential nights at Red-crowned Parrot roosts in McAllen, Brownsville, and Harlingen. The goal of counting on sequential nights was to determine how much variation there was in counts from one night to the next in order to better interpret data from single night counts and develop our long term roost monitoring schedule.
From 7 June – 13 August we conducted a total of 10 pairs of roost counts at the same roost on consecutive nights: Harlingen (N = 5), Brownsville (N = 3) and Mc Allen (N = 2). We calculated the index of daily abundance change as the Absolute Value of ((number on the first night - number on the second night) ÷ (number on the first night)).
To monitor changes at the rate of weeks to months, we also conducted counts separated by approximately two weeks during the same date range (N = 15 pairs of counts at the same roost separated by 14 days). We calculated the index of abundance change over the two-week period as the Absolute Value of ((number on the first night - number on the second night 14 days later) ÷ (number on the first night)).
Counts on sequential nights at the same roost differed by an average of 27 ± 17% (N = 10 pairs of roost counts). As might be expected, the percent change was smaller the larger the total count registered on the first night (Fig. 5, correlation coefficient r = -0.47).
For counts done two weeks apart the change in abundance averaged 62 ± 55% (N = 15 pairs of roost counts). Once again, the percent difference was smaller the larger the total count on the first night (Fig. 6). The variation at the two-week scale was much greater when the roost size was small and dropped to near zero when roosts were large. As a result, this decline could not be modeled with a linear relationship and is modeled here as a decreasing logarithmic function (see Fig. 6 for the equation of the logarithmic function).
The overall variation in roost count values that we see across a broad range of temporal scales from daily to interannual is likely composed of a variety of components: 1) variation in the ability of the observer(s) to detect and accurately count all birds coming to a roost, 2) groups of birds breaking away from the main group and forming small satellite roosts, resulting in more than one roost per city, 3) birds moving among major roosts at the city level, 4) seasonal fluctuations caused by pairs not joining
communal roosts during reproduction, and 5) increases and decreases of true overall population size due to death and reproduction.
To date we have little evidence that the birds break into multiple independent roosts in widely disparate parts of the city. However, that possibility cannot be completely discounted as we have seen birds separate into smaller groups just as they move to the final roosting places and roost a few blocks from each other. However, only during the 5 TPWD counts conducted since January 2016 have we had a protocol by which we would have been able to easily detect the presence of multiple roosting groups. The only other evidence we have is that casual observers have not reported two different roosts on the same night in different sections of the same city.
We know that the other four sources of variation are affecting our nightly roost counts. We have clearly detected evidence of movement among cities in December (#3, see figures 3 and 4 and discussion above) and drops in roost use during breeding (#4, see figures 3 and 4 and discussion above) and with the data we collect over the coming year(s) we should be able to estimate the magnitude of these effects. Given that one of our main interests in this study is to determine the population status for the RCPA in the LRGV, we ultimately would like to determine the change in the total population size from year to year (#5), but this may be challenging given the large number of other sources of variation present in the system. In order to detect changes in the overall population size, we will likely need to use many counts from many different months for each year of the study. These different counts would then need to be compared among years to determine if the population as a whole as increased. The important take-home message of this finding is that it will be difficult to use single annual counts to detect changes in population levels among years. The fact that TPWD is conducting four annual counts per year and plan to do this for the long term is certainly a useful step in our drive to monitor the changes in the RCP population in the LRGV.
The daily variation detected in our study was measured at the level of the individual roost site. It is unlikely that there were major population changes (#5) or large seasonal changes in the number of birds coming to roost (#4) over a period of 24 hours. This suggests that the 27 ± 17% variation from day to day was driven by a mix of variation in detection (#1) and birds moving among cities (#3). The only way to tease apart these two variables will be to conduct simultaneous valley-wide counts at all four major roosts on sequential days.
The variation at the biweekly level (here defined as once every two weeks) is larger than that at the daily level (62% versus 27%). Preliminary statistical analysis suggests that the significance of this value is debatable due in part to the large variation in percent change among nights for both data sets (t-test: P = 0.03, sign rank test: P =0.08, linear regression with roost size on day one as a random variable p = 0.18). However, we strongly suspect that at the biweekly level some of the additional sources of variation are coming in to play. During the time tested (June – August), more birds were returning to the roost post breeding as the chicks fledged. As a result, the 62% likely includes a conglomeration of observer issues (#1), movement among cities (#3), seasonal returns to roost (#4), and true population growth (#5).
Given the very large variation at the biweekly level and our inability to parse out the source of the variation in the counts on consecutive days, we have chosen to monitor the roosts in each city once per week. In this way, we will be able to more closely track the seasonal changes in roost use and the change in percent young at the roosts. In order to better understand the variability in the roosting counts, we are investigating the feasibility of conducting a series of simultaneous counts on sequential nights throughout the LRGV.
Figure 5: Numbers of Red-crowned Parrots roosting in cities across the lower Rio Grande Valley and their percent change on sequential nights. Data are from Harlingen (N = 5), Brownsville (N = 3) and Mc Allen (N = 2) and were collected from 7 June – 13 August. The index of daily abundance change is calculated as the Absolute Value of ((number on the first night - number on the second night) ÷ (number on the first night)). The equation defining the linear regression line and the R2 value are shown on the graph.
Beissinger, Steven R., et al. "Anatomy of a Bottleneck: Diagnosing Factors Limiting Population Growth in the Puerto Rican Parrot." Ecological Monographs 2008: 185. JSTOR Journals. Web. 1 Oct. 2016.
Matuzak, G., and D. J. Brightsmith. 2007. Roosting of Yellow-naped Parrots in Costa Rica: estimating the size and recruitment of threatened populations. Journal of Field Ornithology 78:159-169.
Figure 6: Numbers of Red-crowned Parrots roosting in cities across the lower Rio Grande Valley and their percent change on counts conducted 14 days apart. Data are from Brownsville (N = 5), Mc Allen (N = 4), Harlingen (N = 3), and Weslaco (N = 3) and were collected from 7 June – 13 August. The index of daily abundance change is calculated as the Absolute Value of ((number on the first night - number on the second night) ÷ (number on the first night)). The equation defining the logarithmic regression curve and the R2 value are shown on the graph. .
Protecting Red-crowned Parrots
Red-crowned Parrots (RCPA) can be found in many areas of the Rio Grande Valley. It has been estimated that the population of RCPA is about 1000 in South Texas.
“We suspect the population in South Texas could rival the number found in the wild in Mexico”, said ornithologist Karl Berg, an Assistant Professor in biology at the University of Texas Rio Grande Valley, in a statement.
It is gravely important that our community be aware of the threats that RCPA face and do their part to protect this threatened species. The problem of habitat loss has caused RCPA to move in from Mexico into the United States and we are privileged to have them amongst us.
Below you will find the Harlingen city ordinance prohibiting the capture, kill or injury of a Green Parakeet or a Red-crowned Parrot. Please read the full ordinance for complete details.